27 February 2026

Welcome

Chatting to chatbots could end up being bad for your health, it seems. Moderation in all things? There’s lots of other news too, so let’s get into it.

AI chatbots are driving people crazy

Some Australians are showing signs of psychosis or mania in their interactions with chatbots, according to scientia professor Toby Walsh, chief scientist of the UNSW AI Institute.

As the Guardian reports, OpenAI has said some 560,000 of ChatGPT’s 800 million weekly users have shown signs of these mental health problems, and another 1.2 million have developed potentially unhealthy bonds to the chatbot. Some, of course, are Australians.

“I know because some of them or their loved ones are contacting me by email,” Walsh said.

“They tell me how the chatbot confirms their wild theories. That the chatbot tells them, to quote one email, that they’ve ‘cracked the code’. That they’re ‘the only one that could’.”

The bots are “designed to be sycophantic”, he said.

“They’re designed to confirm what the user says. And they’re designed to draw the user in. They always end with an open question, prompting you to continue the conversation and buy more tokens.”

These comments came from his speech to the National Press Club on Wednesday.

“Social media should have been a wake-up call about the harms of unregulated AI,” Walsh said.

“We’re about to supercharge the sort of harms we saw with social media with an even more powerful and persuasive technology.”

As we reported in December, however, Australia’s National AI Plan has abandoned compulsory guardrails.

Also in the news

• At least 75 Australian businesses have paid off ransomware groups since disclosure of these payments became mandatory for businesses with a turnover over $3 million. Another 19 critical infrastructure providers also paid our ransoms in the same eight-month period. The official advice from the Australian Signals Directorate (ASD) is not to pay.

• The ASD has also released its Azul malware analysis tool as open-source software.

• Former ASD officer and L3 Harris executive Peter Williams has been sentenced to 87 months in jail for selling cyber secrets to Russia.

• The government is kicking off a comprehensive review of Triple Zero legislation after a series of major problems.

• The head of Nine Entertainment wants the government to prioritise the news bargaining incentive, which would force the major online platforms to compensate Australian media for the impact of AI on their reviewed. I mean, of course he would.

• Over at Crikey, Cam Wilson points out that when it comes to Australia’s teen social media ban, we don’t yet know whether it’s working, but that doesn’t seem to be stopping other countries moving ahead. He also covers what we do know so far, and how eSafety’s evaluation process will work.

• For some families, the social media ban has made things harder.

• Australia is getting a new national mobile alert system. It will override device settings to cause everything that has a SIM card to make a loud noise and display a message. Messages can be localised down to a radius of about 260 metres.

• Telstra is retiring ADSL services by 16 November 2027, as well as the old CAN Radio fixed wireless voice systems, something I’d forgotten even existed. CommsDay reports that these services still have about 55,000 and 8,500 customers respectively. Telstra has also stopped rolling out LEOsat links to remote mobile cell sites.

• Almost one-third of public sector organisations are failing to fully disclose AI use, according to the Centre of Excellence for Automated Decision-Making and Society.

• Australia has endorsed the India declaration on AI which calls for further international cooperation on AI across 7 themes: human capital; inclusion for social empowerment: safe and trusted AI; science; democratising AI resources; resilience, innovation and efficiency; and AI for economic development and social good.

• The eSafety Commissioner is looking for young people aged 13 to 24 to join the eSafety Youth Council. Applications close 29 March.

IF YOU’VE BEEN FINDING THIS NEWSLETTER HELPFUL, PLEASE SUPPORT IT: The Weekly Cybers is currently unfunded. It’d be lovely if you threw a few dollars into the tip jar at stilgherrian.com/tip. And thanks to those of you who’ve already done so. Much appreciated.

Elsewhere

• Wikipedia has blacklisted Archive.today after the site’s maintainer made threats, launched a denial of service attack against a blog, and was discovered to be altering the content of archived pages. The site operates under many URLs, including archive.ph, archive.md, and the original archive.is. The FBI is on the case.

• AI firm Anthropic, maker of chatbot Claude, is in a fight with the Pentagon over safety guidelines. As Engadget reports, “ Up until now, the company's core pledge has been to stop training new AI models unless specific safety guidelines can be guaranteed in advance.” I like the Seriously Risky Business headline on this: Is Claude too woke for war? STOP PRESS: Anthropic has rejected the Pentagon’s demands.

• From the Observer, “Meet the vibe lawyers: the AI chatbots wreaking havoc in the justice system”. Ah, self-represented litigants!

• Consulting firm Accenture is deciding who to promote and who to fire based on how often they’re using AI. Using AI? You can stay.

• China is betting cheap AI will get the world hooked on its tech. Meanwhile, Anthropic has accused China of stealing the Claude chatbot, in a process they’ve called “distilling”.

• “Researchers have devised a method that allows large language models (LLMs) to strip anonymity from pseudonymous online accounts at scale for as little as $1.41 per target,” reports iTnews.

• The New York Times has an update (gift link) on that California trial alleging that social media is knowingly addictive and harmful.

• Some philosophers think that, yes, AI really is intelligent.

HOW’S THAT AI BUBBLE GOING? My most recent podcast episode is The 9pm S-Bend of Technology with David Gerard. He’s the editor of the Pivot to AI newsletter, video essay, and podcast. Look for The 9pm Edict in your podcast app.

Inquiries of note

Nothing new this week.

What’s next?

Parliament returns this coming Monday 2 March, with sittings across two weeks, then a one-week gap, then another couple of weeks — including a short week before Easter.

The House of Reps draft legislation program includes debate on the Telecommunications Legislation Amendment (Universal Outdoor Mobile Obligation) Bill, which aims to “ensure that baseline mobile coverage is reasonably available outdoors to all people in Australia on an equitable basis”.

And in the Senate there’s debate on the Freedom of Information Amendment Bill 2025.

As always, the government may add in new legislation to suit their needs.

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.