4 September 2026

Welcome

I have a proposal. I reckon we should stop saying “frontier AI models”, because that’s obviously just spin from an industry trying to frame itself as a bunch of bold pioneers rather than reckless tinkerers. Instead we should say “untested” or “unproven” or “experimental” or the like. Are you with me?

Anyway, the big story this week is the release of some new digital privacy laws, at least in draft form. By sheer coincidence or otherwise, it’s happening at the same time that the risks of smart glasses are in the news. I needn’t remind you of the pushback against the “pervert glasses”.

There’s more, so do read on.

Finally, some draft new digital privacy laws

We’ve been waiting for five years and they’re finally here: the government’s proposed “tranche two” changes to the Privacy Act to update it for the contemporary digital environment.

On Monday the Attorney-General’s Department published an exposure draft of the Privacy Amendment (Personal Data Protection) Bill 2026 and a consultation paper.

Attorney-General Michelle Rowland highlighted four of the changes in her media release, quoted here verbatim:

• a fair and reasonable test, so companies can no longer justify collecting more information than they need or using it for purposes Australians would never expect.

• a right to erasure, allowing individuals to request destroying personal information from large digital platforms like social media and search platforms.

• stronger, more meaningful standards for consent, to ensure Australians are making an informed choice about the information they share, without hidden terms, or take-it-or-leave-it conditions.

• measures to stop businesses from trading in personal information without clear permission, meaning shopping habits, online interests or location data cannot be bought and sold behind Australians’ backs.

You may like to read some more detailed analysis from law firm Corrs Chambers Westgarth.

On the surface there’s a lot to like. The “fair and reasonable” test for data collection is a world first, for example. And there would be a tighter deadline for reporting data breaches to the Office of the Australian Information Commissioner (OAIC): 72 hours rather than the current “as soon as practicable”.

But as usual the details matter, as does enforcement.

“Privacy reform may seem wonkish or a lost cause but, in reality, it’s one of the best tools we have to take back power from big tech and creepy companies,” wrote Digital Rights Watch chair Lizzie O’Shea in the Guardian.

“This proposal is a great first step but it can’t be the only one,” she wrote.

“Well-designed rules mean nothing if they are not enforced. The regulator, the Office of the Australian Information Commissioner, is under-resourced and outmatched in size compared to the corporations it supervises.”

As the eSafety Commissioner’s ongoing legal battles have shown, getting foreign megacorporations to bow to Australian law can be difficult.

The consultation period is short. Submissions close 18 September, which is now just two weeks away.

But what about the smart glasses, you perverts?

Well the short answer is that there won’t be a specific ban. Instead, the A-G put the onus on local government and businesses to implement bans of the technology, “because they can set the terms and conditions of entry into their premises and use of their facilities”.

But being too strict could break the law, according to senior lecturer Mark Giancaspro from Adelaide University.

“Australian businesses can impose whatever terms they like upon consumers, provided those terms are lawful,” Giancaspro wrote in The Conversation.

“Smart glasses — including the best-known Ray Ban Meta glasses — can be fitted with prescription lenses. Alternatively, someone might simply require smart glasses to support their vision. In both cases, a blanket ban might constitute unlawful discrimination under Australia’s Disability Discrimination Act 1992.”

After all, smart glasses have uses other than covert recording — although apparently that one feature could be handy for extremists.

Anyway, it’s probably worth reading the eSafety Commissioner’s view on smart glasses.

WHY NOT SUPPORT THIS NEWSLETTER?: You’re more than welcome to forward these emails to your colleagues, friends, families, and secret intermittent companions. However The Weekly Cybers is currently unfunded. It’d be lovely if you threw a few dollars into the tip jar at stilgherrian.com/tip. Thank you to those of you who’ve already done so.

Also in the news this week

• Attorney-General Michelle Rowland also announced IDLock, described by Cyber Daily as “a new myGov service allowing Australians to block, unblock, and monitor the use of eligible identity documents exposed by data breaches”. It builds on the Credential Protection Register established in 2022, so the government reckons it can start trialling it with a small group of users before the end of this year.

• Some great analysis by the Guardian here. It turns out that “dozens of policy submissions from across the political spectrum incorrectly summarise real research and invent or wrongly cite sources. The hallucination-riddled submissions look convincing enough to end up being cited in subsequent government inquiry reports.

• Also from the Graun, “AI chatbot helps teach online-only psychology classes at Macquarie University”. The university says the chatbot does not replace humans, however, according to Information Age.

• From Mumbrella, “Australia’s media and marketing industry is calling for independent audience measurement of artificial intelligence platforms, including ChatGPT and Gemini, according to the IAB Australia”. After all, how will companies know whether they’re getting good value for what they’re paying to skew AI conversations in their favour? Sorry, to display their ads? Yes, ChatGPT has been accepting ad money for 200 days now and has already billed US$1 billion globally.

• Australia’s News Bargaining Incentive (NBI), which was introduced four months ago, is now under pressure from a group of US Republican politicians.

• From ABC News, “Facial recognition technology ‘scope creep’ is upon us”, a piece focusing mostly on its use in supermarkets and the like.

Another note from TechLeaders 2026

I haven’t had the chance to go through my notes from the TechLeaders 2026 conference, so here’s another story I’ve noticed from there.

• Australia risks becoming a landlord to foreign AI companies, according to Professor Toby Walsh, chief scientist at UNSW’s AI Institute. “Let’s be honest, those data centres are not going to employ very many people once they’re built,” Walsh said. “They’re going to employ loads of people to build them. But most of those profits are going to be for overseas companies [that] don’t pay taxes here.”

Disclosure: Stilgherrian travelled to TeachLeaders 2026 as a guest of Influencing, the organisers.

Elsewhere

• Renown cybersecurity expert Marcus Hutchins has published an essay well worth reading: Machine Speed is a lie: stop trying to fight AI with AI. His argument is that the Cyber Hackers™ have always operated at machine speed, and that most of what you’re hearing is “marketing hype”. As he writes, “If proper proactive defences are deployed, they can cut off pretty much every path attackers will go for. At this point you don’t have to worry about how fast the attacker is moving, because winning is no longer a game of simply outrunning reactive defences. This is where the majority of attackers simply give up and move on to the next target.”

• The glittering allure of AI might be driving the massive growth in datacentres, but of course the planet’s entire digital and communications infrastructure runs on them. They’re vital. Which explains why Russia has told its datacenter operators they need to install defences against drones.

• Robotaxis are trying their luck on London’s medieval roads (New York Times gift link). Presumably companies like Waymo are also employing Londoners to clean out the poop and other nasties from their driverless vehicles.

• The term “digital twin”, at least when it relates to humans, is being used to describe a chatbot that’s been trained on extensive, specific information about an individual. But new research indicates that they contain five key distortions which make them little more than “funhouse mirrors” of the people they are meant to represent. Indeed, they’re only a modest improvement over a generic chatbot.

• “UK AI medical transcription is dangerous and doesn’t work,” reports Pivot to AI.

• Google’s AI overviews reinforce some conspiracy theories, according to new research.

• WIRED journalist Reese Rodgers asked 100 companies for his data, Some deleted it instead. McDonald's, on the other hand, sent him “a stunning 515-page report a few days later that detailed my app interactions in granular detail and predicted I would never stop eating there”.

• Chipmaker Nvidia is reportedly buying Hugging Face for US$12.9 billion. The latter company is “a cloud repository for AI models, similar in some respects to what GitHub is for conventional computer software”, reports Ars Technica, and of course it was also the company hacked by OpenAI’s bots recently.

• I should probably mention the kerfuffle over Lake Ontario following Donald Trump’s executive order to rename it Lake America, and how Google and Apple and some others followed that order even though they’re not part of America’s executive government. There. I’ve mentioned it.

• I should probably also mention that Apple has a new CEO. John Ternus, until now the company’s senior vice president for hardware engineering, replaces Tim Cook after his 15 years in the job.

• Finally, something to enjoy. The Internet Archive has published Vintage Artificial Intelligence, “a curated collection of early computer software claiming some aspect of artificial intelligence as a primary feature, allowing the early interactions of humans and machines before things got a little strange”.

LATEST PODCAST: BRITISH POLITICS AND MORE! The weird goings-on in British politics were a focus of my more recent podcast episode, The 9pm Clacton and On and On with David F Porteous, Scottish author and social researcher. We also discuss Barnaby Joyce’s claim that One Nation is “one of the gayest parties in Australia”, the early days of social media, maths, physics, shrinkflation, and more. Look for The 9pm Edict in your podcast app. NEW EPISODES COMING SOON.

Inquiries of note

Apart from the Privacy Act stuff already mentioned, which of course is very important, there’s nothing new for us this week.

What’s next?

Parliament returns this coming Monday 7 September for two weeks of sittings.

The House of Reps draft legislation program includes debate on the Telecommunications Legislation Amendment (Universal Outdoor Mobile Obligation) Bill.

The Senate draft program includes the Online Safety Amendment (Strengthening Enforcement for the Social Media Minimum Age) Bill 2026.

As always, more can be added on the day according to the government's needs and media cycle desires.

DOES SOMETHING IN THE EMAIL LOOK WRONG? Let me know. If there’s ever a factual error, editing mistake, or confusing typo, it’ll be corrected in the web archives.