Whose voice is it anyway? Designing community resources to navigate AI voice scams

Scams are no longer limited to written messages or suspicious phone calls, with the growing use of synthetic voices built from real recordings, making them harder to detect. 

Today, publicly available voice cloning applications can reproduce a person’s voice from less than a minute of recorded audio. Voice samples can be gathered from social media videos, voice messages or other publicly available recordings and used to generate highly convincing synthetic speech.

While these technologies can have legitimate applications, they are also creating new opportunities for scammers. AI-powered voice scams are emerging as a fast-evolving public threat, combining sophisticated social engineering techniques with increasingly realistic synthetic voices.

To better understand how communities can prepare for these emerging risks, researchers from the ARC Centre of Excellence for Automated Decision-Making and Society worked with Tactical Tech, Good Things Foundation Australia and the Australian Centre for the Moving Image (ACMI) to develop and test a suite of public education resources called AI & You. 

The research explored how AI literacy materials can help people recognise and respond to AI-enabled voice scams. It also highlights the crucial role of mentors, facilitators, and public spaces in enabling AI learning, as these actors are well-positioned to identify and respond to community needs in ways that are contextually relevant and accessible. 

Voice scams exploit trust and intensify existing scam tactics

Voice-based scams are particularly effective because they exploit trust and familiarity, often resulting in distressing and stigmatising experiences for those targeted. 

These scams build on existing scam tactics, amplifying them through scale, speed and realism. Advances in generative AI and machine learning now make it possible to produce highly convincing synthetic voices, while deepfake technologies further enhance the believability of fraudulent scenarios. 

Earlier iterations were easily detectable due to their robotic tone; however, advances in AI and machine learning now enable highly realistic voice cloning that can produce synthetic voices convincingly similar to an original voice.

The hidden cost of scams

Beyond financial loss, scams can have significant social and emotional consequences.

Research has shown that scam experiences are often accompanied by feelings of shame, embarrassment and stigma. Many people are reluctant to discuss what has happened to them or seek support afterwards.

This silence can make the problem worse. It limits opportunities for people to learn from one another’s experiences and can leave those affected feeling isolated.

For this reason, efforts to improve public awareness need to do more than explain how scams work. They must also create spaces where people feel comfortable discussing risks, sharing experiences and seeking help without judgement.

While anyone can be affected by scams, building awareness and confidence in navigating AI-related risks should be tailored to the needs of diverse communities, particularly those who are more digitally excluded or marginalised.

The risks are not equally shared

Anyone can become a victim of a scam. However, some communities face additional challenges

People with limited digital experience, language barriers, lower levels of confidence online, or reduced access to trusted support networks may have fewer opportunities to learn about emerging AI risks or how to report them. Researchers increasingly refer to this as an “AI divide” — a gap not only in access to technology, but also in access to the knowledge, skills and support needed to navigate it safely.

This means public education about AI cannot rely on a one-size-fits-all approach. Resources designed for digitally confident users may not be accessible to older Australians, migrant communities, people with disabilities or those with limited digital literacy.