9 October 2026

Welcome

Data centre operator Firmus was set to become Australia’s biggest company, but has now cancelled its share float. Are investors starting to become wary of the industry’s circular financing? Watch this space.

Meanwhile, Donald Trump says, and I quote: “The White House considers anyone that uses the term, ‘Artificial Intelligence,’ as opposed to the highly accepted new and more accurate term, ‘Super Intelligence,’ THE ENEMY! President DONALD J. TRUMP.”

I am the enemy. And so are you.

OpenAI, Anthropic face parliamentary inquiry

Much was made of the parliamentary inquiry into AI and its public hearings this week, particularly the star appearances by representatives from Anthropic and OpenAI on Tuesday, but did we really learn anything?

OpenAI used AI to write an email — surprise! — but doesn’t know why Australians don’t trust AI. The company’s chief strategy officer Jason Kwon repeated their apology for hacking government websites and promised to do better in the future.

And as ABC News reported, Anthropic’s representatives said their company was “finalising a deal to let Australia's AI Safety Institute independently test its models”.

The Australian headlined the idea from Andrew Forrest’s non-profit Mindaroo of forcing AI giants to pay artists ($), while the AI companied bleated their usual whinge about having to pay for copyright.

But there doesn’t seem to have been much more than that.

To be fair I have yet to read the full Hansard transcript from Tuesday (PDF) or the rest of the hearings and all the other witnesses. Indeed, the fourth and final day is still in progress as I type these words. If anything significant happens I’ll let you know.

Meanwhile the South Australian government has announced the commissioners for its own AI Royal Commission.

OAIC to investigate smart glasses app HeyCyan

The Office of the Australian Information Commissioner (OAIC) has opened an investigation into Shenzhen Qingcheng Future Technology Co Ltd.

According to the OAIC, the company’s HeyCyan app is used by a variety of low-cost smart glasses, ”including Kmart’s Anko branded smart glasses, glasses sold via Big W’s marketplace, and glasses sold via Amazon and other retailers”.

The OAIC has been asking privacy-related questions of various manufacturers and distributors of smart glasses, including Meta, Kmart, Google, and Shenzhen Qingcheng, since mid-August following a request from the attorney-general.

“In the absence of any response from Shenzhen Qingcheng to the OAIC’s preliminary inquiries, I have no assurance that it is protecting personal data of Australians as required by the Privacy Act,” said Privacy Commissioner Carly Kind.

Kind also wrote a blog post providing further background on this issue, including the OAIC’s position on smart glasses and the forthcoming “second tranche” Privacy Act reforms that we discussed last month.

Meanwhile Kmart has removed the Anko glasses from its website.

DON’T WORRY, I’M NOT A BILLIONAIRE: I’m not even particularly well-off. This newsletter is free to read, but it’s unfunded. Crowdfunding also fuels my podcasts The 9pm Edict and Another Untitled Music Podcast. It’d be lovely if you threw a few dollars into the tip jar at stilgherrian.com/tip, or supported this season’s podcast crowdfunder. If you’ve already done so, thank you very much.

Also in the news this week

• Pornhub has returned to Australia, but only for adults with Apple devices. According to BBC News, “Apple users can prove they are over 18 using a credit card, their passport, a driver’s licence or a government-issued ID”. In my case, having had an Apple account for many years seemed to be sufficient.

• As usual, the sex industry is setting the technological pace here. Professor Toby Murray reckons the rest of big tech should follow.

• Data centre operator Firmus has cancelled its share float on the ASX after what ABC News describes as “poor interest from investors amid concerns about the company’s debt and a lack of detail”. That, and perhaps the fact that one of the co-founders has spent a year in jail for insider trading — although to be fair that was 10 years ago. So much for becoming Australia’s biggest company.

• The latest government AI safety thought bubble is that the technology could be regulated like banks and aviation, to “ensure companies are meeting the expectations of safety, without trying to specify every hazard,” according to Andrew Charlton, the assistant minister for science, technology, and the digital economy. “Voluntary” regulations aren’t enough, he said in a speech on Thursday. The buzzword du jour is “systems regulation”.

• CommBank has signed an AI licensing agreement with Copyright Agency, which represents more than 40,000 creators, including publishers, authors, journalists, and visual artists. The licence includes permission to use Australian newspaper text in workplace generative AI tools — but not to train, create, or improve an AI model.

• From Mumbrella, “Melbourne creative agency Kill Boring Dead has offered to pay employees their full salary for two years if they can successfully automate themselves out of a job through the use of artificial intelligence.”

• Subsea cable operator SUBCO’s founder Bevan Slattery is thinking of buying a survey drone and cable-laying ship. The latter would be an expensive bit of kit, but no one in Australia currently owns one.

• Darwin could soon be a major interconnection point for subsea cables.

• The draft online safety legislation that was released last month proposed giving the eSafety Commissioner and approved researchers the power to create sock puppet accounts on social media to test their compliance. Somehow, some people interpreted this to mean that no one else could be anonymous. Those people would be wrong, The Conversation explains. Part of the misunderstanding is down to an unfortunate usage of the term “sock puppet”.

• Former chief scientist Cathy Foley has some advice for policymakers on how to make sense of the tsunami of shitty information.

• Last week the OAIC published new resources on transparency for the use of AI and automated decision-making.

NEW PODCAST: After far too long a break, my good friend Snarky Platypus and I have produced a fresh episode of Another Untitled Music Podcast, AUMP 00008: 2026 So Far. It’s available on that link right there, on Mixcloud, and on YouTube.

Elsewhere

• US secretary of defense war [sic] Pete Hegseth has appointed Elon Musk to lead his war commission, dubbed Project Meridian. He’ll be joined by Palmer Luckey, founder of high-tech defence contractor Anduril, and former Republican House Speaker Newt Gingrich. The Pentagon is worried about potential conflicts of interest, and rightly so.

• The Trump administration has frozen green-card applications for H1-B skilled-worker visas sponsored by Microsoft, Adobe, Cognizant, Infosys, Tata, Wipro, HCL Technologies, and Capgemini. “Our message to Microsoft is: You’re a great American company, but you’ve got to hire great American workers,” said Vice President JD Vance, who leads the government’s anti-fraud task force.

• Apple’s macOS operating system will soon tell you when an AI requests data from a Mac, following allegations that Meta’s Muse AI assistant could access private messages.

• Photos created by AI are getting better, notes the New York Times (gift link) but there are still some tells, as this article explains.

• OpenAI will start watermarking ChatGPT text output by default, but only in the EU.

• From Pivot to AI, the view that Anthropic’s Claude Mythos bot isn’t so much a very very dangerous hacking tool as an expensive PR stunt.

• A North Carolina man has been sentenced to 18 months in jail for AI-assisted streaming fraud, although the US Department of Justice press release calls it “super intelligence” because they have to now. The fraudster had used a network of thousands of bots to stream his AI-created tunes from the major music streaming platforms, sometimes generating more traffic than Taylor Swift’s entire catalog.

• I’ve avoided writing about the massive set of mathematical proofs released by OpenAI the other day because I’m not sure what to make of it. However, cryptographer Matthew Green reckons — sorry for the link to X but that’s where he said it — that the speed at which AI might find new ways to solve problems could mean the end of public-key cryptography, one of the fundamental tools of cybersecurity. Mark Maunder from cybersecurity firm Defiant argues that we’ll therefore need AI maths to keep up with The Bad Guys™.

• SpaceX has more than 11,000 Starlink satellites, accounting for about two-thirds of all active spacecraft in low-Earth orbit. Now the company is calling for better coordination to avoid collisions.

• The Internet Corporation For Assigned Names and Numbers (ICANN) has released the latest instalment of the New Generic Top-Level Domains (gTLD) Program, the list of to-be-approved endings for internet domain names. There’s .adidas and .canva and .kawasaki and .roland and more than 300 other brands — including .taylorswift. And now begins the review process.

• A new BBC Radio 4 documentary explored the links between The Grateful Dead and the birth of Silicon Valley.

Inquiries of note

Nothing new for us this week.

What’s next?

Parliament returns this coming Monday 12 October for four days of sittings.

The Senate draft legislation program includes debate on the Telecommunications Legislation Amendment (Universal Outdoor Mobile Obligation) Bill 2026 and the Telecommunications Amendment (Enhancing Consumer Safeguards) Bill 2025.

As it stands, the House of Reps draft legislation program doesn’t include any debates of interest to this newsletter, but of course that can be changed on the day.

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.