When it comes to protecting personal privacy, opting out is no longer an option.

Smart glasses have upended the idea that consumers still have a choice about the information they’re offering up to tech companies.

Until now, much of our privacy advice assumed that if you didn’t want your personal information captured by consumer technology, you had some ability to opt out – don’t download the app, limit the information you supply, or skip using the service altogether.

AI’s evolution into wearable tech means we’ve reached a point where our privacy depends on other people’s fashion choices.

Authorities are recognising this shift. Some Queensland councils are banning smart glasses at public pools, the Greens want a 12-month pause on their importation and the Norwegian government is moving to ban them altogether.

It’s too late to regulate the thousands of $89 smart glasses that sold out of stores in just a few days last month.

Anyone anywhere could be wearing them on Australian streets, and people are understandably concerned.

That’s because this isn’t just some new gadget, it represents a fundamental change in the privacy social contract.

AI is turning everyday devices into systems that can observe, identify and interpret people without their knowledge.

This is much more significant than ordinary photography because of what the tech can do with what it sees. It can identify, interpret, summarise, remember and connect information.

We need to regulate that capability, rather than simply banning one product.

It was heartening to see Australia’s online regulator moving in this direction with her advice to industry this week.

The eSafety commissioner is calling for makers to ensure the technology automatically blurs people’s faces when filmed by the devices and to make it clearer when filming is taking place.

These are sensible steps because this evolution is just beginning. If it’s not glasses, it will be earbuds, then cameras built into other wearables.

This is ambient AI, where machines can continuously interpret the world around their users.

That means we need rules that focus less on what the device looks like and more on what it does.

Any device that can covertly capture and analyse other people should have to meet certain standards before being sold.

This could look like:

• unmistakable indication when recording is occurring
• automatic blurring of bystanders’ faces
• on-device processing where possible
• minimal retention of incidental footage
• no facial identification of strangers by default
• clear rules about where recording is prohibited
• meaningful penalties when people deliberately use the technology to invade someone’s privacy.

A 12-month import ban, as proposed by the Greens, would buy lawmakers some time and reduce the number of devices entering the country while clear rules are developed.

Importantly, the Greens have also proposed exemptions for legitimate disability uses, because there is no doubt this technology can be a game-changer for those with visual impairments.

The potential benefits are another reason regulation requires careful consideration of the broader change, rather than focusing on just one device.

The federal government has chosen not to include an import ban as it released proposed changes to Australia’s privacy laws this week.

The proposed changes also fail to address some of the key recommendations made by the eSafety commissioner.

The government is spruiking the move to create a “right of erasure”, allowing people to request platforms dispose of any collected personal data.

While welcome, this leaves the onus on the individual, rather than demanding the tech companies take responsibility for how their products perform in the first place.

In comments this week, federal Attorney-General Michelle Rowland also put the onus back on local government and businesses to deal with the implications of products like smart glasses by enacting bans.

That leaves them carrying the cost of monitoring and compliance, rather than ensuring tech companies and manufacturers meet specific requirements.

It’s hardly a major shift from our existing approach to digital privacy, which has largely focused on protecting ourselves.

None of the usual advice around using strong passwords, checking app permissions and taking care when sharing personal information applies to wearable, undetectable AI worn by someone else.

Suddenly a walk around the block has us sharing personal information that can be captured, interpreted, analysed and used by a system we haven’t chosen, without us even knowing.

Our privacy laws need to catch up before that world becomes normal.

Dr James Birt is Associate Professor of Film, Screen and Creative Media at Bond University