Latest
AI + ML

We’ve asked what AI sees, now we should ask what AI hears

Comment: The risk is not that hearing aids simply record what we say. The risk is that AI-enabled hearables learn who we are. The post We’ve asked what AI sees, now we should ask what AI hears appeared first on Newsroom.

TechChallenger Staff3 October 2026 at 16:00 UTC6 min read
We’ve asked what AI sees, now we should ask what AI hears

The risk is not that hearing aids simply record what we say. The risk is that AI-enabled hearables learn who we are.

Dr Grant Searchfield

45 minutes ago02/10/2026

Share on LinkedIn (Opens in new window)

Share on X (Opens in new window)

Share on WhatsApp (Opens in new window)

Share on Bluesky (Opens in new window)

Share on Threads (Opens in new window)

Print (Opens in new window)

Email a link to a friend (Opens in new window)

The next generation of AI-enabled earbuds and hearing aids may do far more than amplify sound. Photo: Getty Images

Expert opinion from University of Auckland - Waipapa Taumata Rau

Comment: Most of us have seen the headlines and concerns about smart glasses that enable wearers to secretly record what they see. As debate intensifies over AI-powered smart glasses capable of discreetly capturing images and video, authorities in Australia and elsewhere are considering, or have already enacted, restrictions in some public settings.

Meta has already signalled awareness of these concerns, previewing an audio-only smart-glasses model without a camera. This raises an intriguing question, if society becomes uncomfortable with devices that can secretly see, how will it respond to devices that can continuously listen?

In a recent University of Auckland study, audiology student Anzel Singh interviewed 18 experts from AI, cybersecurity, hearing healthcare, digital technology and health research. We asked a simple question: what will hearing technologies be capable of by 2030 and 2050?

Foreign Affairs

Rogue AI agents going nuclear? At this election, NZ faces a reality check

Planetary Solutions

AI can fake almost anything, but kids can be taught to spot it

The answers were remarkably consistent. Experts anticipated major health benefits, but they also warned about surveillance, data ownership, behavioural influence and the concentration of power among large technology companies.

The next generation of AI-enabled earbuds and hearing aids (‘hearables’) may do far more than amplify sound. Future devices are being designed to recognise voices, translate conversations, monitor health through in-ear sensors, function as personal assistants, and learn from the acoustic environments around them. Future AI hearing aids and earbuds may not just hear the world around us; they may understand it.

Emerging AI-enabled hearing technology promises enormous benefits. Those interviewed envisioned hearing devices that could identify preferred speakers, provide real-time coaching in social situations, monitor wellbeing, and function as ever-present AI companions.

Interviewees repeatedly suggested that the greatest value of future hearables may come not from sound amplification, but from the vast amount of information they gather.

This could lead to a new era of increasingly personalised health assistance. My research group investigates personalised care for tinnitus (ear and head noise). This new technology is invaluable because we can determine how tinnitus differs from person-to-person and day-to-day making personalised therapy possible. We are working towards a future where we can predict tinnitus and then treat the specific cause – perhaps even before the tinnitus becomes a problem.

In another application a hearing aid could notice subtle changes in speech patterns suggesting cognitive decline months before family members notice, enabling early intervention. Importantly, many of these health benefits can be achieved without storing or transmitting identifiable conversations, highlighting the importance of privacy-by-design approaches.

Interviewees also described a darker scenario: surveillance, manipulation, and the concentration of power among large technology companies. The next generation of AI hearables may continuously monitor speech, identify friends and family, interpret conversations, and build detailed models of users’ health, behaviour, relationships, and daily lives.

Imagine candid conversations being routinely recorded and subject to scrutiny. We only need to look at political ‘hot mic’ incidents to see the potential consequences.

Experts warned that continuously listening devices could collect voice, behavioural and biometric data while building increasingly detailed models of users’ lives. Several noted that speech patterns alone might reveal far more than most people realise, potentially allowing inferences about health, mental wellbeing, social relationships and behaviour.

The hearing aid of 2035 may effectively know who you spend time with, how often you speak, what you talk about, where you had those conversations, and how your communication patterns change over time. These recordings may not be intended for surveillance, but the data may still be retained because they improve device performance.

In the 20th century, most conversations were ephemeral; they disappeared as soon as they were spoken. In the 21st century, conversations have become recordable. The next stage may be that conversations become searchable.

Future AI hearables may capture something far more personal: who we love, who we trust, what worries us, how healthy we are, and how we think. One of the most important concerns emerging from our study was the possibility that AI agents become deeply integrated into users’ lives and increasingly influence behaviour. The future concern may not be eavesdropping. It may be persuasion.

Experts repeatedly imagined AI assistants that are always listening, always available and increasingly personalised. Unlike human relationships, they require no reciprocity. As these systems become more sophisticated, the concern may not simply be what they hear. It may be how effectively they can shape what we think and do.

The question is whether society can establish the guardrails before the technology learns more about us than we know about ourselves.

There is little doubt that AI-enabled hearing technologies have the potential to deliver extraordinary advances in communication, health, and convenience, and New Zealand can help lead these developments. We lead the world in tinnitus research, and these technologies will help us move closer to preventing, predicting, and ultimately, we hope curing some forms of tinnitus.

New Zealand also has an opportunity to help shape the global rules governing the safe and ethical use data recorded by next-generation hearables. Here at the Eisdell Moore Centre at the University of Auckland we recognise the importance of data sovereignty and ownership. Several interviewees argued that indigenous data should not be treated solely as a commercial asset. They suggested Māori concepts of personhood and data sovereignty may offer useful ways of thinking about ownership and protection of the information generated by future AI systems.

Visual recordings through smart glasses are front of mind, and may drive future legislation, but the broader field that includes AI-enabled hearing technologies deserves equal attention. As well as asking, “What can AI see?” We need to ask, “What can AI hear?” and “Who owns what AI hears?”

These devices will create remarkable health benefits because they can understand us. The same capability means they may understand us too well. The long-term issue may not be surveillance but influence. New Zealand has an opportunity to establish the social, ethical, and legal guardrails we need, and which we need now.

Newsroom exists because of readers like you. Help us deliver trusted, independent journalism that serves all New Zealanders. Support our work with a donation today.

Dr Grant Searchfield

Associate Professor Grant Searchfield is based in Audiology, School of Population Health, at the University of Auckland. He is also co-director of the Eisdell Moore Centre for Hearing and Balance Research...

Leave a comment

Cancel reply

Start your day with a curation of our top stories in your inbox

Most Popular

SolarZero tradie welcomes move to hold directors accountable

Zombie towns fight back

Auckland hospo giant bets on waterfront expansion despite surging liquidations

Newsroom daily quiz, Saturday 3 October

The secret diary of .. the campaign: Week 1

Start your day with acuration of our top stories in your inbox.

Facebook Page

Twitter Username

Terms and Conditions

Privacy Policy

New Zealand Media Council

Commenting Rules and FAQ

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Powered by Newspack

Sign in or register

Continue with Google

Enter the code sent to your email

Email address

Enter your password

Sign in by entering the code we sent to , or clicking the magic link in the email.

Resend code

Email me a one-time code instead

Forgot password

Set a password (optional)

I agree to Newsroom's Terms and Conditions. This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Verify your email

We'll send a verification code to %EMAIL%.

Send code

Skip for now

Gift this article

Copy link

More in AI + ML

All AI + ML »