19 June 2026

Welcome

Two international stories kick us off this week, the UK’s planned teen social media ban, and the restrictions on Anthropic’s new Mythos 5 AI model. There’s plenty of other news too.

But we skip over the fact that Elon Musk is a trillionaire — apart from mentioning that Gina Rinehart is all in.

UK launches “Australia plus” teen socials ban

We knew it was coming and here it is. The UK is going to ban social media sites for under-16s. The prime minister, Sir Keir Starmer, used the same phrase that was used heavily here in Australia: “give kids their childhood back”.

Gaming and live-streaming platforms will also be heavily restricted in a teen ban system which seems to go further than any other country so far — although the legislation still needs to be passed.

According to an explainer from the Guardian, “A consultation on online safety closed on 26 May, giving ministers just weeks to come up with policies after receiving more than 116,000 responses. Industry sources and child safety advocates have described the process as ‘rushed’ and driven by a political timeline. It is not clear when the ban could come into force.”

Sounds familiar, no?

Australia’s ban is “floundering”: New York Times

“Six months in, many teens are already back on platforms they were supposed to be blocked from. The ban’s benefits may fall to the next generation,” reports the New York Times (gift link).

They cite eSafety’s own figures that 70% of parents whose kids already had accounts are still on banned platforms.

“So far, these signs of fallibility haven’t deterred other countries that are planning on introducing similar laws. Last month, Britain’s online safety minister, Kanishka Narayan, travelled to Australia to learn about the implementation of the law as his country considers similar steps to protect children,” the NYTimes writes.

US bans export of Anthropic’s Mythos AI

Anthropic has blocked non-US customers from accessing the Mythos 5 and Fable 5 versions of its Clause AI model after the government issued an export control directive on national security grounds.

Mythos 5 is the model supposedly so powerful that it would be too dangerous to release generally. It was already being restricted to friendly governments and other approved customers. Fable is a version with the model’s cybersecurity and biotechnology capabilities blocked.

Apparently the US government was worried that Mythos might be diverted to foreign military intelligence.

Could America’s ban on exporting Mythos be an own goal?

More than 150 cybersecurity leaders have written an open letter calling for the ban to lifted to “maintain America’s lead” in AI cybersecurity technology.

“The Chinese open-weight models are only months behind the best American models, and those are the models we know about. It seems likely that the PRC government has access to private capabilities beyond what has been published,” they write.

“To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous.”

ASPI’s Cyber & Tech Digest has some good coverage of this story. You may also appreciate the analysis at Tech Policy Press.

The latter site also makes a broader observation: “The open internet was always American. Now it's a weapon.”

It’s an arms race, says Hastie, and Australia may fall behind

“The AI race is as dangerous as the nuclear arms race that defined the Cold War,” said Andrew Hastie, shadow minister for industry and sovereign capability, in a speech on Monday.

“Last century, Australia missed the opportunity to become a nuclear power. As a result, we live under the nuclear umbrella of the United States,” he said.

“This century, Australia risks missing the opportunity to become an AI power.”

Hastie also spoke about a “second AI race” between the US and the Peoples’ Republic of China, one which has “geopolitical implications for us all”.

“That second AI race gravitates around Taiwan, the source of 90% of the world’s advanced semi-conductor chips. Taiwan controls the vital AI infrastructure essential to the ambitions of the US and China,” he said.

Also in the news this week

• Australia’s richest person Gina Rinehart’s company has invested $1 billion in SpaceX shares, at least reportedly, and now says Queensland should give some islands to Elon for launch facilities. Reactions are mixed.

• AI infrastructure company Firmus wants to build three datacentres in northern Tasmania which would use 400 megawatts of energy. If they go ahead, and one is already doing so, they might kill the business case for the Bass Strait power cable Marinus Link, which would have exported cheap Tasmania renewable power to the mainland.

• The Tech Policy Design Institute (TPDi) has published a new paper, Expanding AI sovereignty to AI agency. I haven’t had a chance to read it yet, but it looks more than merely interesting.

• A paper from the Australian Strategic Policy Institute (ASPI) argues that “Australia’s National Intelligence Community faces a growing mismatch between how intelligence is produced and how it is now consumed in an AI-shaped information environment”.

• From 1 July you may receive SMS texts labelled as being from “Unverified” and which are diverted into a thread with all the other unverified texts. That’s because they’ll be from an organisation which has failed to register their SMS Sender ID, the branding used instead of a phone number, with the Australian Communications and Media Authority (ACMA).

• The Australian Signals Directorate (ASD) has told organisations they shouldn’t hand software projects to developers who don't have security skills. This straightforward concept is in the latest version of the ASD’s Information Security Manual (ISM).

• ASX Limited (ASX), Australia’s main stock exchange, has admitted to misleading conduct for saying, in February 2022, that its blockchain-based tech replacement project was “progressing well”. (Remember blockchain?) As it happens, the project was scrapped in November that same year, having blown $250 million across seven years. ASX will now be paying a $20.5 million fine.

• Vodafone had a big outage this week. Here’s what went wrong, at least according to one engineering academic.

• From Mumbrella, the observation that AI-generated ads are good enough, at least in terms of efficient production, but they fail to make a proper emotional connection.

• A hacker has used a data breach at the Productivity Commission to abuse journalists via email. As Cyber Daily reports — their writers were among the targets — “the emails contained abnormal content, with the email sent to me [Daniel Croft] having the subject line ‘:(” and simply reading ‘cybercriminals are not terrorists’, as well as a table that featured the words ‘f--- you’.”

END OF FINANCIAL YEAR SUPPORT?: The Weekly Cybers is currently unfunded. It’d be lovely if you threw a few dollars into the tip jar at stilgherrian.com/tip, or just forwarded it to others who might be interested. Thank you to those of you who’ve already done so.

Elsewhere

• Yes, yes, Elon Musk is a trillionaire following the float of SpaceX shares. Enough has been written already. Let’s move on, shall we?

• Estonia, a nation that knows how to use its digitals, is planning to recognise AI agents with digital IDs so we know who’s behind them. “In the future, AI will increasingly carry out digital tasks on our behalf, compiling reports, preparing declarations, or interacting with information systems," said Prime Minister Kristen Michal in a statement. “To that end, it must be clear who is acting on whose behalf with what rights, and who is ultimately responsible.”

• In the US, section 702 of the Foreign Intelligence Surveillance Act (FISA) allowed intelligence agencies to, as the Electronic Frontier Foundation (EFF) puts is, “collect communications from foreigners abroad without a warrant, and routinely sweeps in Americans’ emails, messages, and calls in the process”. This post-9/11 law has now expired.

• China claims “spy sea turtles” are studying its coastline.

• Google's AI Overviews are having trouble distinguishing fact from fiction. In particular, SCP Foundation is a long-running collaborative fan fiction universe, but Google thinks it’s real. Try looking up “SCP-426” for example — but be quick.

• In Korea, the AI and silicon chip boom is inspiring new language. The New York Times explains some of the latest slang in quiz form (gift link).

NEW PODCAST: It’s a personal episode this time. In The 9pm Hometown Visit to Gawler, my good friend Snarky Platypus and I stroll through the South Australian town where I was born and where I lived from age 11 to about 17. Look for The 9pm Edict in your podcast app. It’s a bit different, but you might enjoy it.

Inquiries of note

Nothing new this week.

What’s next?

Parliament returns this coming Monday 22 June. Neither the House of Reps draft legislation program nor the Senate program show anything of particular relevance to this newsletter, but things are often added 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.