17 July 2026

Welcome

I enjoyed my time off in Wollongong last Friday, thanks for asking. And now we’re back. Happy World Emoji Day.

I’ve tried to get excited about the AI “framework”, but it really is just all talk, and about the Telstra outage, but there’s not much to say apart from how you should pay attention to known problems and understand your own networks.

Anyway, I’m pleased to see that mobile phones don’t cause brain cancer.

An AI framework without any actual parts

So many words were written this week about prime minister Anthony Albanese’s plan to establish an Office of AI in his department and to regulate data centres. He used the term “framework”. Some news stories even referred to it as a “blueprint”.

But as Donald Trump might say, it was really just concepts of a plan. Or even just the announcement of the intention to develop some concepts of a plan.

If you can be bothered, you can read Wednesday’s speech and its cringe personal story about Albo working in a bank and explaining ATMs to the punters, and the media release, and discover that this “blueprint” is:

• Some “Australian Standards for AI”, timeline unspecified, and “whole-of-government AI consumer safety priorities in coming weeks”, how many weeks unclear;

• An Office of Artificial Intelligence, to be created “effective today”, though it’s unclear who this is and what it’s meant to do — unless “accelerate implementation of the Australian Standards on a national level” actually means something;

• A National Cabinet meeting in August;

• Something about copyright, because we love our Australian writers and artists and journalists (there are options); and

• Some legislation about all of this “next year”.

It’s obvious to your writer that the government needed an announceable about AI for some reason — probably all the anti-AI and anti-datacentre news lately — and put together this mishmash of motherhood statements.

But apart from giving a few people in the prime minister’s office new business cards, it’s little more than “Oh yeah, we’re going to do something about the things”, most of which were already happening. Kinda. Let’s see what actually happens next.

There is an interesting new report, however...

Also released on Wednesday, the report AI and employment in Australia from the Department of Employment and Workplace Relations. The tl;dr, at least as reported by Information Age, is that AI is not causing mass layoffs.

Telstra’s Big Day Out — of action

OK, so it wasn’t a whole day, but as I’m sure you know on 8 July a timekeeping tech failure at Telstra triggered a nationwide outage that also halted NSW trains and trains across Victoria.

The cause? A known bug in an obsolete server. By failing to replace a 20-year-old box that then cost $20,000, the telco could be hit with $30 million in fines.

These two paragraphs from the Guardian coverage today say it all:

“Telstra said it had made an intentional design change to the equipment to fix an earlier fault — but that this had not been properly documented. This meant maintenance workers arriving at the Melbourne site early last Wednesday were not aware of how the device would be reset,” the Guardian writes.

“A software update had also not been applied to the device, and if that had been done, the outage may not have occurred, Telstra said.”

With the Triple Zero system failing, Telstra then had to conduct hundreds of welfare checks. Some of them were presumably to Senator Sarah Henderson, who was testing the system multiple times. She says she won’t apologise for doing her job.

As this is being written, Telstra executives have been appearing before a Senate committee hearing, so there’s bound to be more news later today.

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 fortnight

• Some 21 medical practices of Adelaide-headquartered Partnered Health suffered a data breach back on 23 June, but only told people this week. They’ve organised an injunction from the Supreme Court of NSW, ordering that the accessed data is not used or published, so that’s fixed then.

• The Australian Broadcasting Corporation (ABC) is running a trial of using AI for journalism using Anthropic’s Claude, although the ABC also uses other AI models for other tasks. One of Claude’s jobs will be turning radio news stories into web articles.

• More than 50% of Australian university assignments used AI.

• Researchers in Queensland took a self-driving Tesla Model Y out for a spin for more than 100 days and recorded “more than 500 safety-critical events where the system required driver intervention or revealed an important limitation in how it interpreted the road environment”.

• Some interesting research from the Australian Institute of Criminology (AIC) on how ransomware victims decide what to do. “Victims rarely tried to negotiate with offenders. The vast majority of victims decided against paying the ransom because they distrusted the offenders,” they write.

MY LATEST PODCAST EPISODE IS ABOUT TRAINS: There’s a strong focus on China and on high-speed trains in The 9pm Not Quite All of the Trains with David Feng, who according to his YouTube channel is an international “rail visionary”. Look for The 9pm Edict in your podcast app.

Elsewhere

• A new lawsuit claims that Meta used AI to sack workers who took sick leave. Or rather, as Reuters puts it, “disproportionately targeted people with disabilities or who took medical leave”.

• Meta has taken down a new Instagram feature that allowed users to create fake images from other people’s content after a few days of massive backlash.

• Mark Zuckerberg admits that AI is not working out the way he imagined.

• Getting an AI chatbot to jump its guardrails could be easier than you think. This is both brilliant and, I think, hilarious.

• After three years of Microsoft 365 Copilot, adoption is under 4.5%, and only 1% of customers use it weekly.

• Along with a social media ban for under-16s, the UK is now talking about a midnight to 6am curfew for 16- and 17-year-olds. “They mostly come at night. Mostly.”

• The European Commission has found that the addictive design features of Instagram and Facebook — things like infinite scroll, autoplay, push notifications, and the platforms’ highly personalised recommender systems — probably breach the EU’s Digital Services Act.

• “Nearly all of the nearly three million papers available on the arXiv preprint server contain details the authors never meant to share,” reports Nature, “from arguments between co-authors and to-do lists acknowledging weaknesses in the text, to passwords, GPS coordinates that can reveal a researcher’s home address, and application programming interface (API) keys.”

• A New York school district is testing robot teachers, which have been made by a company that started out making sexbots. The classroom of the future! I find the photos in both those articles a little disturbing.

• Yet another study shows that there’s no evidence mobile phones cause brain cancer.

• Does your boss resist letting you work from home? They might be a narcissist.

• This headline: “Secretive Silicon Valley startup tries to quash interview, says it’s definitely not working on growing brain-free ‘organ sacks’.”

Inquiries of note

Nothing new for us today.

What’s next?

Parliament is currently on its usual long winter break until 11 August, which is three and a bit weeks away.

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.