Author Kathy Nickels
Date 20 March 2023

ADM+S researcher Prof Dan Angus is co-author on the paper Visual mis/disinformation in journalism and public communications: Current verification practices, challenges, and future opportunities, which has been voted as top paper published in the Q1 journal, Journalism Practice, in 2022-23.

The paper led by Dr TJ Thomson at QUT’s Digital Media Research Centre and co-authored by Prof Daniel Angus, A/Prof Paula Dootson, Dr Edward Hurcombe, and Mr Adam Smith has accrued more than 11,000 views since being published, making it the 14th most-read article in the journal of all time.  

The study provides a state-of-the-art review of current journalistic image verification practices, examines a number of existing and emerging image verification technologies that could be deployed or adapted to aid in this endeavour, and identifies the strengths and limitations of the most promising extant technical approaches. 

Author Kathy Nickels
Date 28 February 2023

ChatGPT is a controversial new language assistant powered by AI. It can write essays, do coding and even structure complex research briefs, all in a matter of seconds.

Launched late November 2022, it now has more than 100 million users according to estimates. 

This new tool, developed by US company OpenAI, is causing concern amongst schools and universities, with fears that students will use the program to write their assignments.

ChatGPT is likely to change the way that students are assessed and force us to rethink what it means to be genuinely creative. 

ADM+S researcher Dr Aaron Snoswell spoke to Athony Funnell on a recent episode of ABC Radio National Future Tense about ChatGPT.

Dr Snoswell suggests that safeguarding and responses to the technology needs to be wide ranging and needs to include Government bodies, experts in the AI industry, system users as well as media.

“Government bodies have a role to play in terms of coming up with regulations, policies, and best practices,” says Dr Snoswell.

Author Kathy Nickels
Date 24 February 2023

Meta, the parent company of Facebook and Instagram has announced it will be testing a new paid verification subscription for users to pay to prove they are real. 

The new offering, called Meta Verified, assigns users a blue verification badge on their profile in exchange for AUD$20 a month.

Professor of Digital Media and Associate Director of the ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S), Jean Burgess spoke to Alex Easton on ABC Southern Queensland Radio about this latest move.

It seems from Meta’s press release that the paid verification is primarily aimed at influencers and content creators, says Prof Burgess.

The paid subscription bundles the blue tick verified badge with other premium features including increased visibility and reach, proactive account monitoring for impersonators, and access to a real person for help with common account issues.

Author Kathy Nickels
Date 22 February 2023

The University of Skövde has appointed Prof Deborah Lupton, its first Honorary Doctorate in the field of Health in the Digital Society.

“It is an amazing feeling to be so honored, particularly as I already have strong connections to and collaborations with colleagues in Sweden and the other Nordic countries. I have always felt very welcome and appreciated in these countries, with lots to talk about in terms of shared interests. This Honorary Doctorate means that I will always have a special relationship with the University of Skövde,” says Professor Lupton.

Prof Lupton is Chief Investigator at the ARC Centre of Excellence for Automated-Decision Making and Society,  node leader of the University of New South Wales, leads the Health focus area and co-leads the People program.

With a background in sociology, as well as media and cultural studies,  Prof Lupton combines qualitative and innovative social research methods with sociocultural theory. Her research focuses on the use of new digital media in medicine and public health. She studies how those media are the focus of increasing interest in society, how they can have unexpected effects for patients and healthcare professionals, and how they can influence how society works with digital technologies in public health and healthcare.

Author Kathy Nickels
Date 21 February 2023

Data collected by Queensland University of Technology node of the ARC Centre of Excellence for Automated Decision-Making & Society (ADM+S) researcher via Twitter’s API appears to support media claims the reach of the tweets of the platform’s billionaire owner Elon Musk have been artificially inflated.

Last week, the tech news site Platformer reported 80 Twitter engineers had been engaged to tweak the platform’s algorithm after Musk noticed a tweet from the US president, Joe Biden, about the Super Bowl outperformed his own, despite Musk having more than three times the number of followers.

The report claimed engineers deployed a new algorithm to artificially inflate Musk’s tweets by a factor of 1,000, ensuring that more than 90% of Musk’s 128.9 million followers would see them. The change reportedly also ensured users who don’t personally follow Musk would see his tweets in their “for you” tab.

Assoc Prof Timothy Graham, Associate Investigator at the Queensland University of Technology node of the ARC Centre of Excellence for Automated Decision-Making & Society (ADM+S) said data he extracted from Twitter using its application program interface appeared to support much of this reporting.

Author Kathy Nickels
Date 21 February 2023

Facebook and Instagram parent company Meta is introducing a paid subscription for users to verify their accounts with a blue tick. 

Meta Platforms has announced it will be testing the monthly subscription service called Meta verified in Australia and New Zealand from this week.

The company says the service will increase the visibility of users’ posts and provide extra protection against impersonation. The move comes after Elon Musk, the owner of Twitter, implemented the premium Twitter Blue subscription back in November. 

Professor of Digital Communication and Chief Investigator at the ARC Centre of Excellence for Automated Decision-making and Society at QUT, Daniel Angus said he doubts that paid verification will make any difference to curb the spread of mis- and dis-information on the platform. 

“We’ve been tracking the spread of this dis-information for many years now, often of a very personal nature … This move will do nothing to actually curb the spread of that.” he said.

“It’s profitable for the platform to maintain pages and groups which spread disinformation. Verifying profiles and extracting more rent from users for doing so is not going to do anything to curb that spread, in fact it may make things worse.”

Authors Kath Albury and Daniel Reeders
Date 30 January 2023

A 2022 survey by the Australian Institute of Criminology found three in four app users surveyed had experienced online abuse or harassment when using dating apps. This included image-based abuse and abusive and threatening messages. A further third experienced in-person or off-app abuse from people they met on apps.

These figures set the scene for a national roundtable convened on Wednesday by Communications Minister Michelle Rowland and Social Services Minister Amanda Rishworth.

Experiences of abuse on apps are strongly gendered and reflect preexisting patterns of marginalisation. Those targeted are typically women and members of LGBTIQA+ communities, while perpetrators are commonly men. People with disabilities, Aboriginal and Torres Strait Islander people, and people from migrant backgrounds report being directly targeted based on their perceived differences.

What do these patterns tell us? That abuse on apps isn’t new or specific to digital technologies. It reflects longstanding trends in offline behaviour. Perpetrators simply exploit the possibilities dating apps offer. With this in mind, how might we begin to solve the problem of abuse on dating apps?

Trying to find solutions

Survivors of app-related abuse and violence say apps have been slow to respond, and have failed to offer meaningful responses. In the past, users have reported abusive behaviours, only to be met with a chatbot. Also, blocking or reporting an abusive user doesn’t automatically reduce in-app violence. It just leaves the abuser free to abuse another person.

Wednesday’s roundtable considered how app-makers can work better with law enforcement agencies to respond to serious and persistent offenders. Although no formal outcomes have been announced, it has been suggested that app users should provide 100 points of identification to verify their profiles.

But this proposal raises privacy concerns. It would create a database of the real-world identities of people in marginalised groups, including LGBTIQA+ communities. If these data were leaked, it could cause untold harm.

Prevention is key

Moreover, even if the profile verification process was bolstered, regulators could still only respond to the most serious cases of harm, and after abuse has already occurred. That’s why prevention is vital when it comes to abuse on dating apps. And this is where research into everyday patterns and understanding of app use adds value.

Often, abuse and harassment are fuelled by stereotypical beliefs about men having a “right” to sexual attention. They also play on widely held assumptions that women, queer people and other marginalised groups do not deserve equal levels of respect and care in all their sexual encounters and relationships – from lifelong partnerships to casual hookups.

In response, app-makers have engaged in PSA-style campaigns seeking to change the culture among their users. For example, Grindr has a long-running “Kindr” campaign that targets sexual racism and fatphobic abuse among the gay, bisexual and trans folk who use the platform.

Author Daniel Angus
Date 17 January 2023

Seeing might not be believing as digital technologies make the fight against misinformation even trickier for embattled social media giants, writes Daniel Angus

In a grainy video, Ukrainian President Volodymyr Zelensky appears to tell his people to lay down their arms and surrender to Russia. The video – quickly debunked by Zelensky – was a deep fake, a digital imitation generated by artificial intelligence (AI) to mimic his voice and facial expressions.

High-profile forgeries like this are just the tip of what is likely to be a far bigger iceberg. There is a digital deception arms race underway, in which AI models are being created that can effectively deceive online audiences, while others are being developed to detect the potentially misleading or deceptive content generated by these same models. With the growing concern regarding AI text plagiarism, one model, Grover, is designed to discern news texts written by a human from articles generated by AI.

As online trickery and misinformation surges, the armour that platforms built against it are being stripped away. Since Elon Musk’s takeover of Twitter, he has trashed its online safety division and, as a result, misinformation is back on the rise.

Musk, like others, looks to technological fixes to solve his problems. He’s already signalled a plan for upping use of AI for Twitter’s content moderation. But this isn’t sustainable nor scalable, and is unlikely to be the silver bullet. Microsoft researcher Tarleton Gillespie suggests: “Automated tools are best used to identify the bulk of the cases, leaving the less obvious or more controversial identifications to human reviewers.”

Some human intervention remains in the automated decision-making systems embraced by news platforms but what shows up in newsfeeds is largely driven by algorithms. Similar tools act as important moderation methods to block inappropriate or illegal content.

The key problem remains that technology ‘fixes’ aren’t perfect and mistakes have consequences. Algorithms sometimes can’t catch harmful content fast enough and can be manipulated into amplifying misinformation. Sometimes an overzealous algorithm can also take down legitimate speech.

Beyond its fallibility, there are core questions about whether these algorithms help or hurt society. The technology can better engage people by tailoring news to align with readers’ interests. But to do so, algorithms feed off a trove of personal data, often accrued without a user’s full understanding.

Authors Alexa Scarlata and Ramon Lobato
Date 11 January 2023

TVs don’t just break down anymore. New problems include apps becoming obsolete and streamers cutting off support for your operating system.

A TV used to be a long-term investment – something you bought knowing it would see you through the next 10 or even 15 years.

Before TVs were ‘smart’, their main function  was to decode signals from broadcast television and from connected devices like DVD players and game consoles. If your TV suddenly stopped working, major hardware faults would hopefully be covered under your manufacturer warranty and statutory guarantees under consumer law.

When you’re buying a new TV today you need to think not just about the quality of the hardware, but about the lifespan of its software

But smart TVs are different because of the complexity of their inbuilt software. When you’re buying a new TV today you need to think not just about the quality of the hardware, but about the lifespan of its software.

This is because your TV’s functionality will change over time. Apps and platforms may not work as well, may refuse to open – or they may disappear altogether.

In this article we explain what you should keep in mind when you buy a smart TV, and what you can do if the functionality of your TV is compromised over time.

What do you get with a smart TV?

Almost all TVs sold in Australia today are smart TVs. This means they can connect to the internet and deliver streaming content via apps.

What many people don’t realise is that when you buy a smart TV you’re locked into using its specific operating system, such as Samsung’s Tizen or LG’s WebOS – just like you’re locked into iOS if you buy an Apple phone or Android if you buy a Samsung.

Each TV operating system works differently, and puts its own spin on interface design, menus, and navigation.

You need to brace yourself for the fact that most apps will eventually stop working on your TV in the years ahead

The operating system also determines the content you can access on your smart TV because it controls the app store. So you need to check before purchase that your preferred TV can run all the apps you might need – not just the big ones like Netflix and YouTube (which are preinstalled on most smart TVs), but also local apps from ABC, SBS, 7, 9 and 10, and any specific movie, sports or gaming apps that you might like to use.

Also, while we all want a smart TV that will work consistently over time, you need to brace yourself for the fact that most apps will eventually stop working on your TV in the years ahead. Even top-of-the-line TVs that cost tens of thousands of dollars are subject to this dreaded phenomenon of ‘app obsolescence’.

The inevitable obsolescence of apps

There are several reasons why apps can become obsolescent.

First, the process of developing and maintaining an app for multiple operating systems and smart TV models is expensive and resource-intensive. Streaming services such as Netflix and iView can see what smart TVs people are using and prioritise particular brands and models accordingly.

In some cases, streaming services will not bother updating apps designed for older TVs, and will instead focus their efforts on newer TVs that are likely to run more smoothly and reach more viewers. As such, they may stop supporting older-model TV apps, or they may remove their apps from particular platforms.

Streaming services will not bother updating apps designed for older TVs, and will instead focus their efforts on newer TVs that are likely to run more smoothly and reach more viewers

This happened in 2019, when Netflix announced that its app would no longer be supported on some Samsung and Panasonic smart TVs purchased in the early 2010s. These devices had “technical limitations” that did not support Netflix’s new digital rights management protocols.

Authors Kathy Nickels
Date 19 December 2022

ADM+S researchers Prof Christine Parker, Prof Jeannie Paterson, Prof Kimberlee Weatherall and colleague Assoc. Prof Paula O’Brien have been awarded funding from the Academy of Social Sciences in Australia (ASSA) Workshops Program to convene leading stakeholders to investigate issues of harmful online advertising and the potential for public interest litigation.

The ASSA Workshops Program for 2023, awarded over $70,000 to convenors from ten different universities to advance research and policy agendas on nationally important issues. 

The ADM+S co-hosted workshop Strategic Public Interest Litigation for Transparency and Accountability of Harmful Digital Marketing: A Researcher-Regulator-Community Dialogue seeks to address challenges of harmful digital advertising. It will bring together key social science and socio legal researchers to investigate predatory and manipulative advertising practices across a range of harmful industries such as alcohol, unhealthy food, and gambling. 

Professor Christine Parker, Chief Investigator at ADM+S, University of Melbourne says that these practices are challenging to investigate. 

“Bringing together scholars, activists and regulators working on these issues in different industries will provide the opportunity to discuss our common challenges.

“We plan to also look at the potential benefits, challenges, and pitfalls of strategic public interest litigation to address these harms.” says Professor Parker. 

Authors Kathy Nickels
Date 16 December 2022

ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S) publications have been named in the APO’s Top Content for 2022 released this week.

The APO’s Top Content for 2022 has listed A Data Capability Framework for the not-for-profit sector and Decentralising data governance: Decentralised Autonomous Organisations (DAOs) as data trusts and as most valuable players (MVPs) – a selection of the most interesting and influential content of the year. 

As part of the Top Content for 2022, the APO also named the Top Ten most clicked resources across 15 broad subject areas for the period December 2021 to November 2022.

The Manifesto for sex-positive social media was listed in both Technology and Communications subject areas and Automated decision making in transport mobilities: review of industry trends and visions for the future was listed in Technology. 

The ADM+S has contributed 17 publications, since joining the APO repository in May this year. 

ADM+S Centre Director, Distinguished Professor Julian Thomas, said “The APO collection has enabled us to communicate our key research work publicly, in detail, in a timely fashion, in a convenient digital format, and in a way which is open to everyone.

The Analysis & Policy Observatory (APO) is one of Australia’s leading open access research repositories. We share APO’s goal of supporting evidence-based policy and public debate on the critical challenges facing Australia, and we’re delighted to be working with APO to make ADM+S research more findable, more useable, and more accessible.”

Authors Kathy Nickels
Date 13 December 2022

ADM+S researcher Arjun Srinivas and colleagues from Kaivalya Plays have received grant funding and support from MediaFutures to produce an interactive theatre performance highlighting the effects of hyper-partisanship and hate speech in India. The performance will particularly focus on the targeting and persecution of minority Muslim women online.

The project  ‘Mining Hate’ will include improvised scenes built from media narratives and content generated by the audience to demonstrate how malicious online actors identify, target and harass victims of online scams and hate speech. Drawing on the Brechtian principle of Verfremdungseffekt, or the “distancing effect”, the live performance seeks to engage audience members in the emotional fallout of being victims.  

Media Futures is a European funded consortium that supports artists to address challenges of disinformation and hate speech in the digital media ecosystem. The recent round of Artists for Media grants were awarded to artists with an innovative artwork concept and production process that critically and materially explores data and technology to question and comment on its impact on individuals and society.

Authors Kathy Nickels
Date 12 December 2022

ABC has published this informative interactive explainer Wrenching open the black box that examines the “black box” of algorithms that are increasingly making decisions about our online and offline lives.

Using relevant examples such as Centrelink’s Robodebt and UK’s visa-processing algorithm, this explainer illustrates how some decision-making systems can be flawed.

The article highlights research and tools in development that can help us understand— and challenge — the decisions that algorithms make about us, as individuals, while others can illuminate bias and discrimination embedded within a system.

The article features:

• Professor Sandra Wachter, Oxford Internet Institute and a tool that generates a number of “nearby possible worlds” that helps to illustrate how different variables (e.g. postcode, gender) could generate different outcomes.

Authors Kathy Nickels
Date 12 December 2022

Last week the eSafety Commissioner released their new Tech Trends Position statement: Recommender systems and algorithms. 

The position statement takes a holistic view of recommender systems that encompasses their benefits and risks, broader uses, and complex interconnected ecosystems.

It is stated that “it is important to assess recommender systems holistically, thinking about their benefits and risks, their range of uses and how they may influence or be influenced by the wider digital environment and socio-political developments”.

This paper provides useful advice for users and guidance for industry. For example it recommends that companies take a more proactive Safety By Design approach to recommender algorithms by considering the risks they may pose at the outset and designing in appropriate guardrails.

 This could include:

• features that allow users to curate how a recommender system applies to them individually and opt out of receiving certain content

Author Kathy Nickels
Date 12 December 2022

East Brisbane State School students have been awarded second place in the National Young ICT Explorers (YICTE) competition 2022.

Leading up to the finals, more than 700 students submitted projects from across Australia to the competition. 

On 10 December, the ARC Centre of Excellence for Automated Decision-Making and Society at QUT hosted the East Brisbane State School Grade 6 students to pitch their project idea at the YICTE virtual finalist event. 

In their project “Runway Racket” the students used an Arduino (a single board microprocessor) with a custom microphone to measure environmental noise in conjunction with the Plane Finder website to identify planes associated with the noise. The monitors were placed in several homes along a flightpath over East Brisbane. 

Mairi McGregor, YICTE 2022 judge, praised the team for their creativity in defining the problem, and accuracy in capturing and measuring data as well as presenting the data. She said that the data and presentation was at a level that could be taken to authorities and companies but also had the potential, with expansion, for commercialisation.

Author Ashwin Nagappa
Date 1 December 2022

The recent rollercoaster of changes to Twitter have inevitably made it the most discussed topic on social media, in our daily conversations, in traditional press and in academia. There are obvious speculations about the future of Twitter. There is also an experience of mass grief and despair for many who benefited from it. And the question looming large is — if not Twitter, where else? This blogpost is not about quitting Twitter or finding a suitable alternative. Many handy resources[1] have already been authored in relation to these issues. Generally, there is a lot of chatter about and on the so-called digital town squares. This has turned attention toward other smaller gathering around street corners and alleyways such as Mastodon. Hence, this blogpost is a brief overview of alternative social media (ASM) platforms (Gehl, 2015) and what it could mean for the future of the Internet or social media as we know them.

The desire for alternative media is not new to social media platforms or the Internet. Before the world wide web (the web) became a commonplace, there were various initiatives across the globe to develop alternatives to dominant broadcast media systems at the time (Rennie, 2006). Community media or alternative media[2] initiatives aimed to create alternative media systems that decentralized decision-making and provided access to the production and circulation of media (Sandoval & Fuchs, 2010). The web[3] had all the capabilities of an alternative media and provided space for user-generated content (Van Dijck, 2009). However, commercial interests transformed the web into a platformized web, where digital platforms became central entities (Helmond, 2015).

Alternative social media (ASM) platforms emerged over a decade ago when commercial social media platforms and platform companies had already established dominance in the digital media ecosystem. ASM platforms aimed to create platforms without advertising revenue and algorithms for content curation or recommendation. And to shift the concentration of power from platform companies into a community of users. However, while the promise and aspirations of ASM platforms invited many users, it was a complicated task for users to operate or govern the platform; especially to understand the nuances of content licensing and the challenges that arose on a large network of users.

The earliest ASMs, such as *diaspora, Twister, and Ello, gained momentary popularity as the Facebook or Twitter killer apps (Zulli et al., 2020, p. 1189). However, they failed to scale up or find viable business or economic models. At the same time, little or no content moderation attracted hate speech or far-right actors to the space. Additionally, the technocratic characteristics of these platforms increased participation barriers (the platform design required new users to learn new technical skills) (Gehl, 2015).

Despite many challenges, ASM platforms did not disappear. Rather, decentralized platforms such as Mastodon were worked on to make them relatively user-friendly. Furthermore, several (open source) ASM projects came together over the years to develop a protocol[4] allowing users to participate across a range of decentralized platforms. This led to the birth of a “Fediverse”, a network of user-run social media platforms. While the “Fediverse has existed since 2018, the recent turn of events have drawn attention to it.

Fediverse (Senst & Kuketz, 2021)

Authors Aaron Snoswell and Jean Burgess
Date 29 November 2022

Earlier this month, Meta announced new AI software called Galactica: “a large language model that can store, combine and reason about scientific knowledge”.

Launched with a public online demo, Galactica lasted only three days before going the way of other AI snafus like Microsoft’s infamous racist chatbot.

The online demo was disabled (though the code for the model is still available for anyone to use), and Meta’s outspoken chief AI scientist complained about the negative public response.

So what was Galactica all about, and what went wrong?

What’s special about Galactica?

Galactica is a language model, a type of AI trained to respond to natural language by repeatedly playing a fill-the-blank word-guessing game.

Most modern language models learn from text scraped from the internet. Galactica also used text from scientific papers uploaded to the (Meta-affiliated) website PapersWithCode. The designers highlighted specialised scientific information like citations, maths, code, chemical structures, and the working-out steps for solving scientific problems.

The preprint paper associated with the project (which is yet to undergo peer review) makes some impressive claims. Galactica apparently outperforms other models at problems like reciting famous equations (“Q: What is Albert Einstein’s famous mass-energy equivalence formula? A: E=mc²”), or predicting the products of chemical reactions (“Q: When sulfuric acid reacts with sodium chloride, what does it produce? A: NaHSO₄ + HCl”).

However, once Galactica was opened up for public experimentation, a deluge of criticism followed. Not only did Galactica reproduce many of the problems of bias and toxicity we have seen in other language models, it also specialised in producing authoritative-sounding scientific nonsense.

Authoritative, but subtly wrong bullshit generator

Galactica’s press release promoted its ability to explain technical scientific papers using general language. However, users quickly noticed that, while the explanations it generates sound authoritative, they are often subtly incorrect, biased, or just plain wrong.

We also asked Galactica to explain technical concepts from our own fields of research. We found it would use all the right buzzwords, but get the actual details wrong – for example, mixing up the details of related but different algorithms.

In practice, Galactica was enabling the generation of misinformation – and this is dangerous precisely because it deploys the tone and structure of authoritative scientific information. If a user already needs to be a subject matter expert in order to check the accuracy of Galactica’s “summaries”, then it has no use as an explanatory tool.

At best, it could provide a fancy autocomplete for people who are already fully competent in the area they’re writing about. At worst, it risks further eroding public trust in scientific research.

A galaxy of deep (science) fakes

Galactica could make it easier for bad actors to mass-produce fake, fraudulent or plagiarised scientific papers. This is to say nothing of exacerbating existing concerns about students using AI systems for plagiarism.

Fake scientific papers are nothing new. However, peer reviewers at academic journals and conferences are already time-poor, and this could make it harder than ever to weed out fake science.

Author Kathy Nickels
Date 20 December 2022

Since Elon Musk purchased Twitter on 27 October he has made a slew of chaotic changes in attempts to raise revenue and to grapple with the complexity of governing a social media platform. 

Although Twitter has only a fraction the amount of users when compared to Facebook, Instagram and WhatsApp, the platform plays a significant role in society and shaping public opinion.

Professor Jean Burgess, Associate Director of the ADM+S Centre says that “Twitter’s unique role is a result of the way it combines personal media use with public debate and discussion.

Authors Kathy Nickels
Date 30 November 2022

In this ABC Radio National Life Matters episode, reporter Nat Tencic explores the relationship between ads, the internet and us.

Nat Tencic does some personal research on advertising on her Twitter, Instagram and Facebook feeds. She said the results “weren’t comforting”. 

In 5 minutes of scrolling on each platform Nat found that on:

• Instagram 28% of story slides were advertisements (12 ads within 43 story slides) 
• Facebook 31% of posts were advertisements (21 ads within 68 posts)
• Twitter 20% ot tweets were advertisements (1 in 5 tweets were promoted)
• Tik Tok 21% of videos were ads (3 ads and 11 regular videos)

Author Kathy Nickels
Date 28 November 2022

Hacks/Hackers recently hosted the winning team of the ARC Centre of Excellence for Automated-Decision Making and Society (ADM+S) Dark Ads Hackathon at ABC Southbank to share their idea for identifying discriminatory patterns in online advertising data.

The idea presented by the multi-disciplinary team has the potential to provide better tools for informing policy, advancing public awareness, and building advocacy for vulnerable groups who are targeted by predatory advertising.

The Hackathon team’s approach draws on data gathered from the Australian Ad Observatory dataset (500,000+ ads donated by 2000 Australian Facebook users) to examine “why am I seeing this?” (WAIST) data, alongside other demographic indicators like income, postcode, and age, to identify discriminatory patterns such as proxy and price discrimination.

Members of Hacks/Hackers and the Hackathon team discussed how the methods could be used to identify advertisement practices across a range of harmful industries such as predatory consumer financial products, alcohol, gambling, and unhealthy foods. The innovative approach and methods developed by the Hackathon team can also be applied to other contexts, for example, to trace illegal advertising practices such as the promotion of vape products to young people.

Authors Daniel Angus and Timothy Graham
Date 22 November 2022

After a few chaotic weeks it’s clear Elon Musk is intent on taking Twitter in a direction that’s at odds with the prevailing cultures of the diverse users who call it home.

Musk has now begun reinstating high-profile users – including Donald Trump and Kanye West – who had been removed for repeated violations of community standards.

This comes off the back of a mass exodus of Twitter staff, including thousands that Musk unceremoniously fired via email. The latest wave of resignations came after an ultimatum from Musk: employees would have to face “extremely hardcore” working conditions (to fix the mess Musk created).

All of this points to a very different experience for users, who are now decamping the platform and heading to alternatives like Mastodon.

So what threats are we likely to see now? And how does one go about leaving Twitter safely?

#TwitterShutDown

With so many experienced staff leaving, users face the very real possibility that Twitter will experience significant and widespread outages in the coming weeks.

Enterprise software experts and Twitter insiders have already been raising alarms that with the World Cup under way, the subsequent increase in traffic – and any rise in opportunistic malicious behaviour – may be enough for Twitter to grind to a halt.

Aside from the site going dark, there are also risks user data could be breached in a cyberattack while the usual defences are down. Twitter was exposed in a massive cyberattack in August this year. A hacker was able to extract the personal details, including phone numbers and email addresses, of 5.4 million users.

One would be forgiven for thinking that such scenarios are impossible. However, common lore in the technology community is that the internet is held together by chewing gum and duct tape.

The apps, platforms and systems we interact with every day, particularly those with audiences in the millions or billions, may give the impression of being highly sophisticated. But the truth is we’re often riding on the edge of chaos.

Building and maintaining large-scale social software is like building a boat, on the open water, while being attacked by sharks. Keeping such software systems afloat requires designing teams that can work together to bail enough water out, while others reinforce the hull, and some look out for incoming threats.

To stretch the boat metaphor, Musk has just fired the software developers who knew where the nails and hammers are kept, the team tasked with deploying the shark bait, and the lookouts on the masts.

Can his already stretched and imperilled workforce plug the holes fast enough to keep the ship from sinking?

We’re likely to find out in the coming weeks. If Twitter does manage to stay afloat, the credit more than likely goes to many of the now ex-staff for building a robust system that a skeleton crew can maintain.

Hate speech and misinformation are back

Despite Twitter’s claims that hate speech is being “mitigated”, our analysis suggests it’s on the rise. And we’re not the only researchers observing an uptick in hate speech.

The graph below shows the number of tweets per hour containing hate speech terms over a two-week period. Using a peer-reviewed hate speech lexicon, we tracked the volume of 15 hateful terms and observed a clear increase after Musk’s acquisition.

Misinformation is also on the rise. Following Musk’s swift changes to blue tick verification, the site tumbled into chaos with a surge of parody accounts and misleading tweets. In response, he issued yet another stream-of-consciousness policy edict to remedy the previous ones.

Author Lauren Hayden
Date 10 November 2022

Digital advertising is microtargeted, ephemeral, and unobservable. Ads such as those seen on social media are shown only to select users based on behavioural, demographic, and psychographic data the platform has been able to collect about them. These ads may be published for a limited time, often less than 24 hours, and they are invisible from view after they expire.

This is referred to as ‘dark advertising’. Researchers, advocates, and governments have little ability to monitor online advertising and are therefore unable to hold advertisers accountable for potentially harmful practices such as targeting underage users with ads for alcohol or gambling.

The ADM+S Centre’s Tech for Good: Dark Ads Hackathon challenged teams to create a ‘pretotype’ solution for monitoring, analysing and studying dark advertising. The three-day event, held at RMIT in Melbourne from 28 – 30 September 2022, hosted attendees from across Australia and an array of presenters representing consumer advocacy organisations, research groups and the tech industry.

Although teams were working towards solutions, the Hackathon generated big questions that are only the beginning of the conversation.

Big Question #1 – How does dark advertising affect users?

Existing research shows that digital advertising has been used to target at-risk groups with excessive messaging around harmful products such as alcohol, gambling and unhealthy foods. The ‘dark’ nature of advertising hinders the ability to monitor and report these harmful targeting patterns.

As the first panel discussion highlighted, dark advertising is more than just predatory targeting. Advertisers have the power to artificially limit consumer choice, exploit dynamic pricing for optimal potential revenues, and employ dark patterns to nudge shopping behaviours. User data is sold to third parties which allows for further microtargeting and behavioural manipulation.

Dark advertising is shorthand for a broader, automated consumer culture which affects us all.

Author Timothy Graham
Date 7 November 2022

Following Elon Musk’s acquisition of Twitter on October 27, the world’s richest man proposed a range of controversial changes to the platform. With mounting evidence that he is making it up as he goes along, these proposals are tweeted out in a stream-of-consciousness manner from Musk’s Twitter account.

Primarily to raise revenue, one of the ideas was to charge US$8 a month to obtain a verified status – that is, the coveted blue tick badge next to the account handle.

Within the space of a few days, the paid verification change has already been rolled out in several countries, including Australia, under the Twitter Blue subscription service.

More than just verification

According to Twitter, the blue tick lets people know an account of interest is authentic. Currently, there are seven categories of “public interest accounts”, such as government office accounts, news organisations and journalists, and influencers.

Yet this seemingly innocuous little blue icon is far from a simple verification tool in Twitter’s fight against impersonation and fraud.

In the public view, a verified status signifies social importance. It is a coveted status symbol to which users aspire, in large part because Twitter’s approval process has made it difficult to obtain.

That’s partly because the blue tick has a controversial history. After receiving widespread condemnation for verifying white supremacists in 2017, Twitter halted its verification process for more than three years.

There’s a fundamental mismatch between what Twitter wants the blue tick to mean versus how the public perceives it, something the Twitter Safety team itself acknowledged in 2017.

But they didn’t resolve it. When Twitter resumed verifying accounts systematically in 2021, it wasn’t long until the process began to fail again, with blue ticks being handed out to bots and fake accounts.

Moreover, the public is still confused about what the blue tick signifies, and views it as a status symbol.

Lords and peasants

Musk’s stream-of-consciousness policy proposals may reflect his own preference for interacting with verified accounts. Despite his repeated claims of “power to the people” and breaking the “lords and peasants” system of verified versus non-verified accounts, I ran a data analysis of 1,493 of Musk’s tweets during 2022, and found that more than half (57%) of his interactions were with verified accounts.

Evidently, having a verified status makes one worthy of his attention. Thus, Musk himself arguably views the blue tick as a status symbol, like everyone else (except Twitter).

However, Musk’s US$8 blue tick proposal is not only misguided but, ironically, likely to produce even more inauthenticity and harm on the platform.

A fatal flaw stems from the fact that “payment verification” is not, in fact, verification.

Fact from fraud

Although Twitter’s verification system is by no means perfect and is far from transparent, it did at least aspire to the kinds of verification practices journalists and researchers use to distinguish fact from fiction, and authenticity from fraud. It takes time and effort. You can’t just buy it.

Despite its flaws, the verification process largely succeeded in rooting out a sizable chunk of illegitimate activity on the platform, and highlighted notable accounts in the public interest. In contrast, Musk’s payment verification only verifies that a person has US$8.

Author Kathy Nickels
Date 31 October 2022

The winning idea from the Tech for Good: ADM+S Dark Ads Hackathon proposes new methods to identify online advertising practices that could involve price discrimination.

Professor Daniel Angus, Associate Investigator at the ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S Centre) said the winning idea presents a particularly powerful technique for uncovering forms of misleading and discriminatory advertising.

“This technique will mean that identifying some forms of dark advertising practices will no longer be like finding a needle in a haystack,” said Professor Angus.

The Tech for Good: ADM+S Dark Ads Hackathon 2.5 day event brought together over 40 participants from social science, humanities, and computer science to hack new ideas and methods for better transparency in online advertising.

“The diversity of ideas and potential for impact was extraordinary. While large technology firms continue to drag the chain on advertising accountability, it was refreshing to see our participants offer new ideas and approaches to these significant issues.” said Professor Angus.

The Hackathon was hosted by the ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S) in collaboration with government and consumer rights organisations, who recognised an urgent need for better transparency and accountability  following recent examples of price discriminiation, scam advertising and predatory targeting in online advertising spaces.

The winning pitch, Using postcodes to identify discriminatory patterns in online advertising data,  used the existing ADM+S Australian Ad Observatory database of half a million advertisements donated by close to 2,000 participants alongside statistical data associated with postcodes to identify patterns of price discrimination based on userlocation. 

The team also suggested building a visual interface to help both researchers and consumers quickly identify discrimination and other unethical  advertising practices.

Other ideas presented at the hackathon included: 

• Gamifying private and ethical ad donation
• Empowering users to protect themselves from harmful advertising online
• Reducing the amount of harmful body image advertising experienced on social media
• Helping social media users capture and report Dark Ads
• Dance steps to understand and visualise personal advertising algo-rhythms

Author Kathy Nickels
Date 11 October 2022

ADM+S PhD candidate Dominique Carlon (QUT) has been announced as the winner of the inaugural ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S) Higher Degree Research (HDR) Student Essay Prize.

Higher Degree Research students from the ADM+S were invited to submit a 2,000 word essay to challenge existing perspectives or suggest new directions of research in automated decision-making (ADM) in the field of news and media.

Carlon’s winning essay “Bots are more than human” argues that debates about the risks and benefits of bots having human-like qualities overlooks other creative and interesting possibilities that they could offer society. 

Dr James Meese (Co-leader of the News and Media Focus area at the ADM+S Centre, RMIT) said that the  essay was genuinely innovative.

“The essay will no doubt inspire academia and industry to think more deeply about how to best deploy bot technologies in the future” said Dr Meese.

Author Kathy Nickels
Date 6 October 2022

The ADM+S Dark Ads public panel brought together government representatives, consumer rights organisations and researchers from the ARC Centre of Excellence for Automated Decision-Making and Society to discuss key issues in online advertising.

Panel experts discussed concerns about unregulated online advertising practices with examples of predatory advertising, price discrimination, and scam ads and how these practices impact vulnerable consumers. 

The panelists agreed that advertising is becoming harder than ever before to hold accountable and that there is an urgent need for better online advertising transparency and accountability.

Associate Professor Nicholas Carah (ADM+S, UQ) moderated the discussion on the key issues in online advertising.

“We can’t see the ads [that are being delivered online] and this is a concern as advertising plays such a fundamental role in shaping our public life” said Associate Professor Nicholas Carah

“And for some categories we have real questions and concerns for vulnerable consumers and harmful products that we need to be able to address collectively”. 

During the discussion, questions were raised on whether the expanding hyper-personalisation of online advertising still fits within the traditional definition of advertising. 

Authors Kate Mannell and James Meese
Date 9 September 2022

Doomscrolling can be a normal reaction to living through uncertain times. It’s natural to want to understand dramatic events unfolding around you and to seek out information when you’re afraid. But becoming absorbed in bad news for too long can be detrimental.

A newly published study has found that people with high levels of problematic news consumption are also more likely to have worse mental and physical health. So what can you do about it?

We spoke to Australians in the state of Victoria about their lengthy lockdown experiences and found how they managed to stop doomscrolling. Here are some tips to help you do the same.

Doomscrolling – unhelpful and harmful

“Doomscrolling” describes what happens when someone continues to consume negative news and information online, including on social media. There is increasing evidence that this kind of overconsumption of bad news may have negative impacts.

Research suggests doomscrolling during crises is unhelpful and even harmful. During the early COVID-19 pandemic, consuming a lot of news made people feel overwhelmed. One study found people who consumed more news about the pandemic were also more anxious about it.

Research into earlier crises, like 9/11 and the Boston Marathon bombings, also found that sustained exposure to news about catastrophes is linked to negative mental health outcomes.

Choosing to take control

During the peak of COVID-19 spread, many found themselves doomscrolling. There was lots of bad news and, for many people, lots more spare time. Several studies, including our own, have found that limiting news exposure helped people to cope.

Melbourne, the state capital of Victoria, experienced some of the longest-running lockdowns in the world. Wanting to know how Victorians were managing their news consumption during this time, we launched a survey and held interviews with people who limited news consumption for their own wellbeing.

We found that many people increased their news consumption when the lockdowns began. However, most of our participants gradually introduced strategies to curb their doomscrolling because they realised it was making them feel anxious or angry, and distracted from daily tasks.

Our research found these news-reduction strategies were highly beneficial. People reported feeling less stressed and found it easier to connect with others. Here are some of their strategies, which you might want to try.

1. Make a set time to check news

Rather than checking news periodically across the day, set aside a specific time and consider what time of day is going to have the most positive impacts for you.

One participant would check the news while waiting for her morning cup of tea to brew, as this set a time limit on her scrolling. Other participants preferred saving their news engagement for later in the day so that they could start their morning being settled and focused.

Authors Nicholas Carah, Aimee Brownbill, Amy Shields Dobson, Brady Robards, Daniel Angus, Kiah Hawker, Lauren Hayden and Xue Ying Tan
Date 7 September 2022

Once upon a time, most advertisements were public. If we wanted to see what advertisers were doing, we could easily find it – on TV, in newspapers and magazines, and on billboards around the city.

This meant governments, civil society and citizens could keep advertisers in check, especially when they advertised products that might be harmful – such as alcohol, tobacco, gambling, pharmaceuticals, financial services or unhealthy food.

However, the rise of online ads has led to a kind of “dark advertising”. Ads are often only visible to their intended targets, they disappear moments after they have been seen, and no one except the platforms knows how, when, where or why the ads appear.

In a new study conducted for the Foundation for Alcohol Research and Education (FARE), we audited the advertising transparency of seven major digital platforms. The results were grim: none of the platforms are transparent enough for the public to understand what advertising they publish, and how it is targeted.

Why does transparency matter?

Dark ads on digital platforms shape public life. They have been used to spread political falsehoods, target racial groups, and perpetuate gender bias.

Dark advertising on digital platforms is also a problem when it comes to addictive and harmful products such as alcohol, gambling and unhealthy food.

In a recent study with VicHealth, we found age-restricted products such as alcohol and gambling were targeted to people under the age of 18 on digital platforms. At present, however, there is no way to systematically monitor what kinds of alcohol and gambling advertisements children are seeing.

Advertisements are optimised to drive engagement, such as through clicks or purchases, and target people who are the most likely to engage. For example, people identified as high-volume alcohol consumers will likely receive more alcohol ads.

This optimisation can have extreme results. A study by the Foundation for Alcohol Research and Education (FARE) and Cancer Council WA found one user received 107 advertisements for alcohol products on Facebook and Instagram in a single hour on a Friday night in April 2020.

How transparent is advertising on digital platforms?

We evaluated the transparency of advertising on major digital platforms – Facebook, Instagram, Google search, YouTube, Twitter, Snapchat and TikTok – by asking the following nine questions:

• is there a comprehensive and permanent archive of all the ads published on the platform?
• can the archive be accessed using an application programming interface (API)?
• is there a public searchable dashboard that is updated in real time?
• are ads stored in the archive permanently?
• can we access deleted advertisements?
• can we download the ads for analysis?
• are we able to see what types of users the ad targeted?
• how much did it cost to run the advertisement?
• can we tell how many people the advertisement reached?

Author Kathy Nickels and Zahra Stardust
Date 31 August 2022

Social media platforms have the capacity to shape broader attitudes towards sex and nudity through decisions about what can and can’t be posted online. And yet, it will be of no surprise that the community standards and content moderation practices of dominant social media platforms are not very sex positive.

Platforms currently make private, arbitrary and unaccountable decisions about the kinds of sex and sexualities that are visible in online spaces. These decisions have seen plus-sized profiles, top surgery fundraisers and chestfeeding people flagged for “excessive nudity” and “sexual solicitation”.

Platforms have pre-emptively shut down spaces that have been safe havens for systemically marginalised communities and actively shadowbanned, demoted, de-monetised, suspended and deplatformed groups as diverse as sex workers, people of colour, LGBTQIA+ folk, disabled people, fat activists, women and sex educators.

Current trends in regulation create a hostile environment for those for whom sex is an active, visible part of life, especially when state legislation incentivises platforms to remove all sexual content.

Seeking positive change to social media policies, ADM+S researcher Dr Zahra Stardust and fellow academic and community experts brought together the voices of stakeholders and advocates to develop the Manifesto for Sex Positive Social Media.

The Manifesto originated from a community lab on Alternative Frameworks for Sexual Content Moderation hosted at the 2021 RightCon Summit on Human Rights in the Digital Age where the group considered how social media platforms could better understand sexual content, communication, expression, and representation.

Dr Stardust is a socio-legal scholar working at the intersections of sexuality, technology, law and social justice.

Dr Stardust says “We believe platforms can learn from a long history of sex-positive thinking. Taking a sex positive approach to social media will require structural changes to the current assemblage of power, labour and value, including contesting existing practices of surveillance capitalism, online gentrification and sexual stigmatisation.

Author Kathy Nickels
Date 22 August 2022

Transport systems and vehicles are rapidly becoming automated, often in ways invisible to us. Cars have increasingly automated features like auto brake, cruise control and auto parking. Ride sharing, and car or bike sharing services which are managed through automated digital platforms are increasingly popular. We plan and navigate our routes using data driven recommendations delivered by smartphone apps or in-car systems.

It is predicted that in the near future fully self-driving cars will be on the roads, flying cars and drones will be in the skies, and that we will be using automated mobility systems (such as Mobility as a Service) that will create seamless travel experiences for us by connecting different modes of transport in one journey.

But how are the automated technologies behind these systems being developed? How are the future people who will be using these systems being envisioned? And what kinds of participation are needed to ensure trusted and safe futures for all?

Join discussions on these questions and more at the Future Automated Mobilities: Towards Hope, Justice, and Care symposium being hosted by the ARC Centre of Excellence on Automated Decision-Making and Society (ADM+S) 20 – 21 October 2022.

Attendees can join in-person at RMIT University in Melbourne, Australia or online. All activities are free, to secure your spot visit the event registration page.

Key issues to be discussed at this event:

• Why do we need to be careful? Where are the limitations, flaws and inequities, in dominant anticipatory visions and narratives? What could possibly go wrong if we place ADM at the centre of a predictive, automated and data-driven mobilities future?
• How might we anticipate trusted mobilities futures based in principles of care? And what could go right if we do? What might more realistic, ethical and responsible automated mobilities futures might look like? What does safety actually entail when we explore it from the perspective of the diverse groups of people who might encounter ADM in transport mobilities?
• How can attending to these questions help us move us forward into more hopeful mobility futures? What are the shared values that might make this possible? Where and with whom should we be imagining, designing and testing our future automated mobilities? And what should be our first steps to achieve this collectively and collaboratively?

Author Kathy Nickels
Date 17 August 2022

The Queensland Academy of Arts and Sciences (QAAS) council has elected ADM+S researchers Professor Daniel Angus, Professor Axel Bruns and ADM+S International Advisory Board Member Distinguished Professor Stuart Cunningham AM to join the ranks of the academy.

Invited membership to the QAAS recognises significant eminence and contributions to the arts and sciences as a practitioner.

Professor Daniel Angus says that the QAAS is doing immensely important work in bringing together thought leaders from a range of disciplines.

“Having straddled the social and computing sciences for most of my career I know how important it is to have mentors who can support inquiry from a range of theoretical and methodological backgrounds.

I am honoured to receive this recognition and look forward to continuing to grow the reputation and recognition of the arts and sciences in Queensland” said Professor Angus.

Professor Daniel Angus is Professor of Digital Communication in the Digital Media Research Centre (DMRC) at QUT and an Associate Investigator at the ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S).

Daniel’s research focuses on the development and application of visual computational analysis methods in communication and media studies, with a specific focus on conversation and social media data.

Daniel has been involved in computer science research for 20 years and he contributes regularly to media and industry on the impact of technology on society.

Author Kathy Nickels
Date 8 August 2022

The Tech for Good: ADM+S Dark Ads Hackathon public to be held in Melbourne from 28 – 30 September provides an opportunity for participants to work with consumer rights organisations and academic experts to hack new ideas and concepts for ensuring better accountability for online advertising.

Participants will create conceptual designs and pitches that will contribute to new research into the ways that online advertising practices can be more transparent and social media companies accountable.

The event is being led by researchers from the Australian Research Council’s Centre of Excellence for Automated Decision-Making and Society (ADM+S) with support from Victorian Health Promotion Foundation, ABC, the consumer advocacy group CHOICE, the Consumer Policy Research Centre (CPRC) and the Foundation for Alcohol Research and Education (FARE).

Professor Mark Andrejevic, researcher from the ADM+S and organiser of the hackathon, says that the problem of false, discriminatory, and predatory advertising online is a serious one that requires innovative strategies for providing transparency and accountability.

“The aim of the Dark Ads Hackathon is to involve participants not just from coding and tech backgrounds but also from social sciences and humanities.

“We need to a multi-disciplinary approach to develop some measure of accountability for online targeted advertising” said Professor Andrejevic

A panel of consumer rights and industry experts will inspire participants to ideate new approaches to better online advertising. Speakers include:

• Kate Bower, Consumer Data Advocate (CHOICE)
• Dr Aimee Brownbill, Senior Policy and Research Advisor, Foundation for Alcohol Research and Education (FARE)

Author Natalie Campbell
Date 29 July 2022

On 28 July Telstra launched its Reconciliation Action Plan for 2022-2025, committing significant investments in remote infrastructure and digital inclusion programs, including the ARC Centre of Excellence for Automated Decision-Making and Society’s (ADM+S) Mapping the Digital Gap project which aims to improve digital inclusion outcomes and access to services in remote Aboriginal and Torres Strait Islander communities.

The Reconciliation Action Plan endeavours to create an inclusive Australia where Aboriginal and Torres Strait Islander peoples are connected digitally and empowered to thrive in those contextual environments. Telstra’s partnership with ADM+S is one step toward making this happen.

Led by Dr Daniel Featherstone, ADM+S researchers partnered with 12 remote first nations communities across South Australia, Queensland, Western Australia, New South Wales and the Northern Territory.

Author Kathy Nickels
Date 26 July 2022

ADM+S researcher Dr Anjalee de Silva has been awarded a Women’s Leadership Institute Australia Fellowship for her work on creating tangible change for gender equality in Australia.

Women’s Leadership Institute Australia Fellowships are awarded to those who are leaders in their respective fields, women who have innovative approaches and the courage, conviction and capacity to create real change.

Dr Anjalee de Silva’s work at the ADM+S Centre and the University of Melbourne, Melbourne Law School examines vilification or ‘hate speech’ directed at and about women, as well as the role of law in deterring, regulating, and mitigating the harms of such speech.

Author Lauren Kate Kelly
Date 18 July 2022

Online grocery shopping has boomed since the pandemic began in 2020, with Woolworths and Coles steadily expanding their home-delivery offerings. Rapid delivery is the latest frontier.

Woolworths and Coles Express have been offering on-demand deliveries through UberEats and Doordash since last year. Woolworths recently launched the Metro60 app which promises home delivery within an hour to select suburbs.

These arrangements have received little fanfare, yet they signal a significant shift for supermarket workers.

As part of ongoing research, I study how the gig economy is transforming conditions of work within traditional employment. To find out how interacting with delivery platforms affects supermarket employees, I interviewed 16 experienced “personal shoppers” at Woolworths and Coles who fill delivery orders from supermarket shelves.

The labour of on-demand grocery

In supermarkets that offer on-demand home delivery, the work of the personal shopper takes on a faster pace. For Woolworths employees, for instance, an UberEats order can drop in at any time, setting off an alarm until the order is accepted and picking begins. As one personal shopper explains:

We get this weird dinging sound that everyone dreads. You have to pick that order within the half hour or within the hour … it can drop in at any time. So if you’re sitting there having lunch for an hour, you still have to go do it because you’ve got that KPI to hit.

All the (scanner) guns in the store drop that sound. So it reverberates through the store. The customers can’t hear it because they don’t know what it is. But all of us know what it is.

The on-demand orders must be prioritised alongside existing orders, requiring the personal shopper to juggle competing time crunches simultaneously.

It’s urgent, and they just pop out of nowhere. So you don’t really know when they’re coming until they’re there. It’s super stressful. I dislike them immensely.

Enter the gig worker

Once the order is picked from the supermarket aisles, the employee hands it over to a gig worker for home delivery. Supermarket staff say their interactions are brief and often impersonal.

It’s a complete mess. You have no idea who’s coming to pick up these things. And it’s just people showing up with their headphones in showing you that they’ve got this order on their phone. There’s no real rhyme or reason to any of it.

For supermarket workers, gig workers are neither colleagues nor customers, yet they play an essential role in home delivery and customer service.

When things go awry, however – such as a missing bag or broken eggs – it’s the supermarket staff who field those complaints. Similarly, when personal shoppers run behind schedule it has punitive flow-on effects for gig workers.

The on-demand model may, by design or otherwise, pit two groups of workers against each other, fostering frustrations at both ends.

Most of the time they’re pretty good. They deal with it. It’s just those bad times where we might be behind and then they don’t deal with it very well.

Author Kathy Nickels
Date 15 July 2022

Researchers investigating whether people’s search results vary depending on who they are will be analysing over 350 million search results to uncover the impact of the recommendation algorithms producing Google and YouTube search results.

The Australian Search Experience invited Australians to install a browser plugin that automatically queried Google Search, Google News, Google Video, and YouTube several times a day on key topics. 

Over the past 12 months, the plugin searched on our participants’ Chrome, Edge, and Firefox browsers on 48 topics, from “who should I vote for in the federal election” to “COVID vaccines” and “Tokyo Olympics”.

These data donations from the more than 1,000 citizen scientists who had installed the plugin were crucial to developing a broader perspective on the search results that ordinary Australians encounter as they use these search engines.

Chief Investigator Professor Axel Bruns, an internationally renowned Internet researcher in QUT’s Digital Media Research Centre, said the project explores whether search engines have the potential to create ‘filter bubbles’ or to promote misinformation and disinformation.

“These results will be analysed to understand the personalisation of search results for critical news and information, across key platforms including Google and YouTube, based on the profiles that these platforms establish for their different users.

Author Kathy Nickels
Date 12 July 2022

Dating apps are continually under pressure from civil society, media and governments to address safety concerns resulting from the use of these platforms. In response to pressures dating apps have added a number of so-called safety features. But how safe do these features actually make their users?

In this recent article Surveillance does not equal safety: Police, data and consent on dating apps published in Crime, Media, Culture: An International Journal, authors Dr Zahra Stardust, Dr Rosalie Gillett and Prof Kath Albury draw on empirical accounts of app use – and popular media reporting – to problematise commonsense assumptions about dating apps, safety, technology, policing and surveillance.

The authors use a critical criminological perspective as a lens to think about accountability by transforming the systems under which consent is navigated, alongside a public health approach that demonstrates the value of adopting a nuanced and contextual approach to gender and sexual diversity.

The article raises concerns that some of these safety features actually increase surveillance as user’s data is shared with external ‘consent apps’ and law enforcement agencies. This shared data has the potential to make users less safe – and this is particularly the case for app users who are marginalised or stigmatised on the basis of their race, sexuality, gender, health status, employment or disability.

Author Kathy Nickels
Date 11 July 2022

Is it possible to design and govern ethical sex tech at scale? What might intersectional, public interest sex tech look like?

Dr Zahra Stardust, Dr Jenny Kennedy and Prof Kath Albury have released a new report ‘Public Interest Sex Tech Hackathon: Speculative futures and participatory design’ which captures the research findings, provocations, challenges and insights from the ADM+S Public Interest Sex Tech Hackathon held earlier this year.

“The report makes a unique contribution to public conversations on public interest technologies, ethical data governance and design justice,” said Dr Stardust.

The Public Interest Sex Tech Hackathon brought together designers, technologists and communities to workshop ideas of how sexual technologies can be designed and governed in ways that prioritise public interest benefit.

Participants explored the ethical potential of sex tech for safety, pleasure and health and worked with industry mentors including Samantha Floreani from Digital Rights Watch and Eliza Sorensen from Assembly Four, a collective of sex workers and technologists.

Participants were invited to create open-source designs and pitches that would contribute to new research into the ways that ‘big data’ can be used for sexual and reproductive health, wellbeing, rights and justice.

Author Mark Andrejevic and Gavin Smith
Date 27 June 2022

Private companies and public authorities are quietly using facial recognition systems around Australia.

Despite the growing use of this controversial technology, there is little in the way of specific regulations and guidelines to govern its use.

Spying on shoppers

We were reminded of this fact recently when consumer advocates at CHOICE revealed that major retailers in Australia are using the technology to identify people claimed to be thieves and troublemakers.

There is no dispute about the goal of reducing harm and theft. But there is also little transparency about how this technology is being used.

CHOICE found that most people have no idea their faces are being scanned and matched to stored images in a database. Nor do they know how these databases are created, how accurate they are, and how secure the data they collect is.

As CHOICE discovered, the notification to customers is inadequate. It comes in the form of small, hard-to-notice signs in some cases. In others, the use of the technology is announced in online notices rarely read by customers.

The companies clearly don’t want to draw attention to their use of the technology or to account for how it is being deployed.

Police are eager

Something similar is happening with the use of the technology by Australian police. Police in New South Wales, for example, have embarked on a “low-volume” trial of a nationwide face-recognition database. This trial took place despite the fact that the enabling legislation for the national database has not yet been passed.

In South Australia, controversy over Adelaide’s plans to upgrade its CCTV system with face-recognition capability led the city council to vote not to purchase the necessary software. The council has also asked South Australia Police not to use face-recognition technology until legislation is in place to govern its use.

However, SA Police have indicated an interest in using the technology.

In a public statement, the police described the technology as a potentially useful tool for criminal investigations. The statement also noted:

There is no legislative restriction on the use of facial recognition technology in South Australia for investigations.

A controversial tool

Adelaide City Council’s call for regulation is a necessary response to the expanding use of automated facial recognition.

This is a powerful technology that promises to fundamentally change our experience of privacy and anonymity. There is already a large gap between the amount of personal information collected about us every day and our own knowledge of how this information is being used, and facial recognition will only make the gap bigger.

Recent events suggest a reluctance on the part of retail outlets and public authorities alike to publicise their use of the technology.

Authors Megan Richardson, Jake Goldenfein and Mark Andrejevic 
Date 22 June 2022

In January 2020, the New York Times published an exposé on Clearview AI, a facial recognition company that scrapes images from across the web to produce a searchable database of biometric templates.

A user uploads a ‘probe image’ of any person, and the tool retrieves additional images of that person from around the web by comparing the biometric template of the probe with the database.

It soon emerged that law enforcement agencies around the world, including Australian law enforcement, were using Clearview AI, often without oversight or accountability. A global slew of litigation has followed, challenging the company’s aggregation of images, creation of biometric templates and biometric identification services.

Findings against the company mean that Clearview AI no longer offers its services in certain jurisdictions, including the UK, Australia and some US states.

But this has hardly stopped the company. Clearview is expanding beyond law enforcement into new markets with products such as Clearview Consent, offering facial recognition to commercial entities that require user verification (like finance, banking, airlines and other digital services).

Other companies are selling similar facial recognition products to the general public. It’s clear the private sector has an appetite for these technical capacities: CHOICE has exposed how numerous Australian retailers build and deploy facial recognition for various purposes, including security and loss prevention.

Facial recognition and data privacy laws

The OAIC investigated Clearview AI in partnership with the UK Information Commissioner’s Office, and determined in November last year that the company breached several of the Australian Privacy Principles (APPs) in the Privacy Act 1998. The Clearview AI determination is the most extensive consideration of facial recognition by the Office of the Australian Information Commissioner (OAIC) to date.

In Australia, the deployment of facial recognition is primarily governed by data privacy (or data protection) legislation that, while designed for different technologies, has been somewhat successful in constraining the use of facial recognition tools by private companies and Australian law enforcement.

“The OAIC investigated Clearview … and determined that the company breached several of the Australian Privacy Principles”

But a closer look shows an awkward relationship between facial recognition and existing privacy law.

While the Australian Information Commissioner should be applauded for interpreting Australian law in ways that framed the use of facial recognition by Australian companies as privacy violations, it is unclear whether the reasoning would be applicable in other facial recognition applications, or would survive further examination by the courts if appealed.

In fact, the specifics of how facial recognition works, and is being used, challenge some of the basic ideas and functions of data privacy law.

Commission’s finding reveals complex issues
A few aspects of the determination highlight the complexities in the relationship between data privacy law and facial recognition.

For Clearview AI to be subject to the APPs, it has to process “personal information”. This means information about an identified or reasonably identifiable individual.

Clearview argued the images it collects by scraping the web are not identified, and that the biometric information it created from those images was not for the sake of identification, but rather to distinguish the people in photos from each other. It acknowledged that providing URLs alongside images may help with identification, but not always.

The Commissioner, however, found that because those images contained faces they were ‘about’ an individual. They were also reasonably identifiable because biometric identification is the fundamental service that Clearview provides. The biometric information (the templates) that Clearview extracted from those images was deemed personal information for similar reasons.

“Collecting sensitive information without consent is permissible in cases of threat to life or public safety – those exceptions did not apply in this case”

After finding that Clearview was processing personal information, the Commissioner evaluated whether Clearview was also processing “sensitive information”, which is subject to stricter collection and processing requirements. That category includes “biometric information that is to be used for the purpose of automated biometric verification or biometric identification; or biometric templates”.

The latter is generally understood to be the data collected for the sake of enrolment in a biometric identity system. Collection and processing of sensitive information is typically prohibited without consent under APP 3.3, subject to very narrow exceptions in APP 3.4.

While the Commissioner acknowledged that collecting sensitive information without consent is permissible in cases of threat to life or public safety, those APP 3.4 exceptions did not apply in this case.

Covert image collection and the risk of harm  

The Commissioner was particularly concerned with this type of “covert collection” of images because it carried significant risks of harm. For instance, it created risks of misidentification by law enforcement, being identified for purposes other than law enforcement purposes, and created the perception for individuals that they are under constant surveillance.

Such collection was not lawful and “fair” (as required by APP 3.5) because:

• the individuals whose personal and sensitive information was being collected by Clearview would not have been aware or had any reasonable expectation that their images would be scraped and held in that database
• the information was sensitive
• only a very small fraction of individuals included in the database would ever have any interaction with law enforcement
• although it did have some character of public purpose – being a service used by law enforcement agencies – ultimately the collection was for commercial purposes.

Is a facial image ‘personal information’?

The Clearview AI case is a fairly egregious example of data processing in breach of the law. But the legal status of other facial recognition tools and applications is less clear.

For instance, it is not always clear that facial images collected will constitute personal information. A debate continues in other jurisdictions about the status of facial images and the scope of data protection law. The European Data Protection Board, for instance, was silent on this point in its 2019 advice on video information processing, and there have been conflicting interpretations in the literature.

“A debate continues in other jurisdictions about the status of facial images and the scope of data protection law”

Although there does appear to be some consensus that images from photography or video that are recorded and retained in material form should be considered personal information, this still requires that the person in an image be reasonably identifiable.

That was not controversial in the Clearview case because images were collected for the sake of building a biometric identification system, and its database of images and templates was retained indefinitely.

Authors Zofia Bednarz, Kayleen Manwaring and Kimberlee Weatherall
Date 20 June 2022

What if your insurer was tracking your online data to price your car insurance? Seems far-fetched, right?

Yet there is predictive value in the digital traces we leave online. And insurers may use data collection and analytics tools to find our data and use it to price insurance services.

For instance, some studies have found a correlation between whether an individual uses an Apple or Android phone and their likelihood of exhibiting certain personality traits.

In one example, US insurance broker Jerry analysed the driving behaviour of some 20,000 people to conclude Android users are safer drivers than iPhone users. What’s stopping insurers from referring to such reports to price their insurance?

Our latest research shows Australian consumers have no real control over how data about them, and posted by them, might be collected and used by insurers.

Looking at several examples from customer loyalty schemes and social media, we found insurers can access vast amounts of consumer data under Australia’s weak privacy laws.

Your data is already out there

Insurers are already using big data to price consumer insurance through personalised pricing, according to evidence gathered by industry regulators in the United Kingdom, European Union and United States.

Consumers often “agree” to all kinds of data collection and privacy policies, such as those used in loyalty schemes (who doesn’t like freebies?) and by social media companies. But they have no control over how their data are used once it’s handed over.

There are far-reaching inferences that can be drawn from data collected through loyalty programs and social media platforms – and these may be uncomfortable, or even highly sensitive.

Researchers using data analytics and machine learning have claimed to build models that can guess a person’s sexual orientation from pictures of their face, or their suicidal tendencies from posts on Twitter.

Think about all the details revealed from a grocery shopping history alone: diet, household size, addictions, health conditions and social background, among others. In the case of social media, a user’s posts, pictures, likes, and links to various groups can be used to draw a precise picture of that individual.

What’s more is Australia has a Consumer Data Right which already requires banks to share consumers’ banking data (at the consumer’s request) with another bank or app, such as to access a new service or offer.

The regime is actively being expanded to other parts of the economy including the energy sector, with the idea being competitors could use information on energy usage to make competitive offers.

The Consumer Data Right is advertised as empowering for consumers – enabling access to new services and offers, and providing people with choice, convenience and control over their data.

In practice, however, it means insurance firms accredited under the program can require you to share your banking data in exchange for insurance services.

The previous Coalition government also proposed “open finance”, which would expand the Consumer Data Right to include access to your insurance and superannuation data. This hasn’t happened yet, but it’s likely the new Albanese government will look into it.

Author Jarni Blakkarly
Date 15 June 2022

Major Australian retailers Kmart, Bunnings and The Good Guys are using facial recognition technology in stores, raising concerns among privacy experts.

The use of this developing technology, which captures and stores unique biometric information such as facial features (known as a ‘faceprint’), would come as news to most customers.

In a recent inquiry, CHOICE asked 25 leading Australian retailers whether they use facial recognition technology, and analysed their privacy policies. Based on the policies and the responses they received, Kmart, Bunnings and The Good Guys appear to be the only three that are capturing the biometric data of their customers.

Privacy policies not easy to find

“Most of these privacy policies you have to search for online, and they’re often not easy to find,” says CHOICE consumer data advocate Kate Bower. “But because we’re talking about in-person retail shops, it’s likely that no one is reading a privacy policy before they go into a store.”

CHOICE staff members also visited some of these stores in person as part of the investigation.

Bower says the Kmart and Bunnings stores they visited had physical signs at the store entrances informing customers about the use of the technology, but the signs were small, inconspicuous and would have been missed by most shoppers.

The collection of biometric data in such a manner may be in breach of the Privacy Act.

Facial recognition on the rise

Mark Andrejevic, professor of media studies at Monash University and a member of the ARC Centre of Excellence for Automated Decision-Making and Society, tells CHOICE that the use of facial recognition by retailers is in its early stages in Australia. But he predicts it will increase as the technology becomes cheaper and more effective.

We don’t have a clear set of regulations or guidelines on the appropriate use of the technology

“The first concern is notice and consent, it’s not in highly visible forms of public notification that would invite people to understand what’s taking place,” says Andrejevic.

“I think the other set of concerns is we don’t have a clear set of regulations or guidelines on the appropriate use of the technology. That leaves it pretty wide open. Stores may be using it for the purposes of security now, but down the road, they may also include terms of use that would say that they can use it for marketing purposes.”

Authors Aaron Snoswell, Henry Fraser and Rhyle Simcock
Date 25 May 2022

The first serious accident involving a self-driving car in Australia occurred in March this year. A pedestrian suffered life-threatening injuries when hit by a Tesla Model 3 in “autopilot” mode.

In the US, the highway safety regulator is investigating a series of accidents where Teslas on autopilot crashed into first-responder vehicles with flashing lights during traffic stops.

The decision-making processes of “self-driving” cars are often opaque and unpredictable (even to their manufacturers), so it can be hard to determine who should be held accountable for incidents such as these. However, the growing field of “explainable AI” may help provide some answers.

Who is responsible when self-driving cars crash?

While self-driving cars are new, they are still machines made and sold by manufacturers. When they cause harm, we should ask whether the manufacturer (or software developer) has met their safety responsibilities.

Modern negligence law comes from the famous case of Donoghue v Stevenson, where a woman discovered a decomposing snail in her bottle of ginger beer. The manufacturer was found negligent, not because he was expected to directly predict or control the behaviour of snails, but because his bottling process was unsafe.

By this logic, manufacturers and developers of AI-based systems like self-driving cars may not be able to foresee and control everything the “autonomous” system does, but they can take measures to reduce risks. If their risk management, testing, audits and monitoring practices are not good enough, they should be held accountable.

How much risk management is enough?

The difficult question will be “How much care and how much risk management is enough?” In complex software, it is impossible to test for every bug in advance. How will developers and manufacturers know when to stop?

Fortunately, courts, regulators and technical standards bodies have experience in setting standards of care and responsibility for risky but useful activities.

Standards could be very exacting, like the European Union’s draft AI regulation, which requires risks to be reduced “as far as possible” without regard to cost. Or they may be more like Australian negligence law, which permits less stringent management for less likely or less severe risks, or where risk management would reduce the overall benefit of the risky activity.

Legal cases will be complicated by AI opacity

Once we have a clear standard for risks, we need a way to enforce it. One approach could be to give a regulator powers to impose penalties (as the ACCC does in competition cases, for example).

Individuals harmed by AI systems must also be able to sue. In cases involving self-driving cars, lawsuits against manufacturers will be particularly important.

However, for such lawsuits to be effective, courts will need to understand in detail the processes and technical parameters of the AI systems.

Manufacturers often prefer not to reveal such details for commercial reasons. But courts already have procedures to balance commercial interests with an appropriate amount of disclosure to facilitate litigation.

A greater challenge may arise when AI systems themselves are opaque “black boxes”. For example, Tesla’s autopilot functionality relies on “deep neural networks”, a popular type of AI system in which even the developers can never be entirely sure how or why it arrives at a given outcome.