The Weekly Cybers #94
Twitch added to the social media age restrictions just 19 days before they start, platforms to get a Digital Duty of Care, courts overwhelmed with AI garbage, and much more.
21 November 2025
Welcome
Prepare for the deluge of news about the social media age restrictions — to just from me but from everyone. One way or another, an enormous number of Australians will be affected.
In a somewhat parallel move, the government is also introducing a Digital City of Care, which means reducing the risks for everyone, not just kids.
And it goes without saying that there’s a whole bunch of stories that involve AI. It’s inevitable.
Against this background, I’m pleased to see that we’re starting to get more nuanced and analytical pieces and the ole of social media and AI in our society. I expect more as we near the end of the year — just 34 sleeps until Christmas! — and writers of all kinds become more introspective.
Social media age restrictions: 19 days to go and Twitch gets added to the list — but not Pinterest
As the magic date of 10 December approaches, social media platforms are preparing for the age restrictions — and the eSafety Commissioner is still fiddling with the list of who’s included and who’s not.
On Friday the commissioner added Twitch to the list. The platform is mainly used by gamers who livestream their action and chat with viewers.
“Following Twitch’s own self-assessment, eSafety assessed Twitch as meeting the criteria for ‘age-restricted social media platform’, because it has the sole or significant purpose of online social interaction with features designed to encourage user interaction, including through livestreaming content,” says the press release.
“eSafety has also informed Pinterest it does not consider it subject to age restrictions on the basis it does not currently meet the criteria for an ‘age-restricted social media platform’.”
Remember, though, it’s not up to the eSafety Commissioner to determine these matters, but the communications minister and, ultimately, the courts. Watch this space for the inevitable legal battles.
Facebook and Instagram start warning under-16s
Meta’s platforms Facebook and Instagram have started sending warning messages to users they think are under 16, using SMS, email, and in-app messaging.
The restrictions also apply to Threads, since that platform requires an Instagram account to use.
“Meta will begin stopping access to existing accounts and blocking under-16s from registering new accounts from 4 December, with access removed for all affected accounts by 10 December, the company said,” the Guardian reports.
Affected users can download their content, and can choose to either return to the account when they turn 16, or just delete it.
Roblox is first to roll out age-guessing technology
Roblox is rolling out age-guessing face scanning to get ahead of age restrictions in Australia and then elsewhere. They plan to limit chats to similar age groups.
As the Guardian reports, “Users will be placed into the following groups: under nine; nine to 12; 13 to 15; 16 to 17; 18 to 20; or 21 and over. Children will be able to chat only with others in their age group and similar ones. For example, a child with an estimated age of 12 will be able to chat only with under-16s. Images and video used for the checks would not be stored, Roblox said.”
(I have added semicolons to that quote. The Guardian’s semicolon phobia made that sentence almost unreadable.)
However as The Conversation discussed, Roblox will need to do more to keep kids safe, given its extensive problems, even though all users wanting to use the chat features will now need to show their face to be guessed — or provide an ID.
So much news via Cam Wilson at Crikey!
Friend of the Cybers Cam Wilson continues to do excellent work on the age restrictions over behind the Crikey paywall.
This week he’s learned that the government is already tracking teenagers’ experiences to see what difference the restrictions will make.
“Efforts have already begun to collect data on what young people know and think about the ban, on compliance and circumvention, and on their well-being to establish a baseline that will be used to assess the program,” he writes.
And on Friday he reported that a high Court challenge will be filed by a group called the Digital Freedom Project. They say the ban will harm LGBTIQ+ teens, among others.
Curiously, the group is being assisted by Katherine Deves, the former Liberal Party candidate for Warringah, who has a history of anti-trans comments.
Government to introduce a Digital Duty of Care
A new legislated Digital Duty of Care will place the onus on digital platforms to “proactively keep Australians safe and better prevent online harms”, said the government last Thursday — yes somehow I missed it.
“The duty of care will put the onus on industry to prevent online harms at a systemic level, instead of individuals having to ‘enter at their own risk’,” said communications minister Michelle Rowland.
Platforms will need to “continually identify and mitigate potential risks, as technology and service offerings change and evolve”.
Your writer imagines that the relevant legislation will be introduced to parliament in the coming week.
However there should be a more nuanced view of online safety, according to Sunita Bose, managing director of Digital Industry Group Inc (DIGI), an industry group representing the big platforms.
“Often the companies that are in the spotlight are the ones investing the most,” she told the Commercial Disco podcast.
“They’re the most mature when it comes to the really challenging questions of how you protect people online.”
LATEST PODCAST: Space archaeologist Dr Alice Gorman aka Dr Space Junk, astrophysicist Rami Mandow, and I discuss the 25th anniversary of the International Space Station’s habitation earlier this month, the colonisation of space, and much more in The 9pm Offworld Colonies with Dr Alice Gorman and Rami Mandow. Look for “The 9pm Edict” in your podcast app.
Choose-your-own-rules is the future of content moderation
by Cam Wilson, first published in The Sizzle, reprinted with permission.
TikTok will soon let you choose to see less AI content (Engadget). Using Content Credentials metadata to identify what’s AI slop, TikTok will let users “dial things down” by choice. It’s a baby step towards the approach of Bluesky — which also announced some new moderation policies to allow more granular content reporting and account violations (Engadget) — which even allows users to sign up to additional voluntary moderation, like the Blacksky community (WIRED).
The Sizzle: Content moderation is one of the most complicated, fraught and least rewarding parts of running a social media platform. While I think there should be a clear baseline of what platforms will and won’t allow, I also think that handing more of the control for this to people is a good thing. This already happens in a de facto way via the recommendation engines, but I think making it more explicit and under a user’s conscious control can only be a good thing.
Also in the news
- High Court chief justice Stephen Gageler says the use of AI in Australian courts has reached an “unsustainable phase”, with judges having to act as “as human filters and human adjudicators of competing machine-generated or machine-enhanced arguments”.
- The government is “walking away” from a gambling ad ban on social media, which is turning out to be an unpopular move.
- Data obtained under freedom of information shows that Australia’s mining and manufacturing sectors take months to detect cyber breaches.
- From ABC News, “A series of cyber attacks on defence supply chain contractors has compromised material relating to Australia’s weapons programs. Information relating to the ADF’s $7 billion Redback infantry fighting vehicle contract was among the details allegedly leaked”. Ah, this takes me back to 2017 when I broke the news that secret F-35, P-8, and C-130 data was stolen from an Adelaide engineering firm. I didn’t know it was an Adelaide firm until then defence industry minister Christopher Pyne blurted it out in a press conference.
- Cybersecurity academic Dr Jill Spay has been appointed to review critical infrastructure laws, the Security of Critical Infrastructure (SOCI) Act 2018. That act was amended by the Cyber Security Act 2024 last year.
- TPG Telecom has confirmed a customer in Sydney died last week after outdated software on an older Samsung phone could not make Triple Zero calls. TPG should have done more, say experts.
- Optus Mobile has paid a penalty of $826,320 for failing to comply with telco anti-scam rules.
- Australia joined the US and UK in sanctioning a pair of Russians who were accused of operating services for cybercriminals.
- Ethics adviser and academic Ed Santow, a former human rights commissioner, has spoken out against AI “slop” in a landmark speech. There’s much more to his arguments than this brief paragraph suggests.
- The QUT Digital Media Research Centre has updated the papers at Digital Media Demystified, with explainers covering the social media age restrictions, the importance of Australian-made media, and dealing with disinformation.
- In the digital age, Australia’s Indian Ocean Territories matter again, because that’s where the fibres run.
- The Department of Finance’s new government AI review committee, due to be established mid-2026, will be dominated by public servants, even though plenty in the industry were previously asked for their advice.
- GovAI, a core part of the government’s AI plan announced last week, is no silver bullet for legacy IT, according to Olivia Shen, director of the strategic technologies program at the United States Studies Centre.
- “Australian businesses have actually been slow to adopt AI,” apparently, though surely “slow” is a relative term.
IF YOU FIND 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. Please consider.
Elsewhere
- China’s open-source AI model DeepSeek has now surpassed OpenAI’s GPT series in the app store download charts, triggering some OMFG CHINA fears. “Behind the cultivated image of easy access innovation lies a troubling reality of risks that may threaten security, privacy, national sovereignty, and societal stability,” wrote Jonathan Ping for the Lowy Institute’s The Interpreter. Meanwhile, Seriously Risky Business writes that AI-powered espionage will favour China. Only last week, US company Anthropic announced that it had detected the use of agentic AI in a “highly sophisticated” espionage campaign out of China.
- An AI-powered stuffed animal was pulled from the market after disturbing interactions with children.
- Elon Musk’s AI chatbot Grok continues to deliver the weirdness, saying that Musk could drink piss better than any human in history. Meanwhile his Grokipedia is full of white nationalist talking points and racial pseudoscience, something which France is investigating.
- From Aeon, “Huge swathes of human knowledge are missing from the internet. By definition, generative AI is shockingly ignorant too.”
- Meta has won a long-running monopoly case brought by the US Federal Trade Commission (FTC) by convincing the judge that “personal social networking” as a business category is dead. With everyone watching videos, Meta is now in competition with entertainment apps such as TikTok and YouTube.
- Oracle did a $300 billion deal with OpenAI, and it's already looking very bad indeed.
- In cryptocurrency news, can trades on the public blockchain be illegal if the computer code allows them? If the code allows it, then it’s legal, right?
- Elizabeth Sandifer, author of Neoreaction a Basilisk: Essays on and Around the Alt-Right, has posted an interesting essay on AI and the Turing Test.
- Australia’s eSafety Commissioner, Julie Inman Grant, has been asked to give evidence about Australia’s laws to a US congressional committee after having been described as a “noted zealot for global takedowns” who “threatens speech of American citizens”.
- More than 57,000 primary schools in the Philippines will get a new cybersafety guide for kids thanks to the Australian Federal Police translating the picture book into Tagalog.
- “The next privacy battleground is inside your brain,” writes the New York Times (gift link). Indeed, what are the privacy implications of brain monitoring or even Neuralink-style brain implants?
WHY NOT LISTEN TO OUR PODCAST ABOUT MUSIC? My good friend Snarky Platypus and I produce a music podcast, Another Untitled Music Podcast. Look for it in your podcast app. The latest one may make you very angry.
Inquiries of note
Nothing new this week.
What’s next?
Parliament returns this coming Monday 24 November for what is currently scheduled to be the last sitting week of 2025.
As usual the draft legislation program for the Senate lists the debates of bills already in play, including the catchy Communications Legislation Amendment (Australian Content Requirement for Subscription
Video On Demand (Streaming) Services) Bill 2025. I love that this bill has nested parentheses in the title.
The draft program for the House of Reps includes the video streaming bill, which they need to finish to hand off to the Senate, the Telecommunications Amendment (Enhancing Consumer Safeguards) Bill 2025, and the Copyright Amendment Bill 2025.
As always the government may, and almost certainly will, introduce new bills as and when they see fit.
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.
The Weekly Cybers is a personal weekly digest of what the Australian government has been saying and doing in the digital and cyber realms, on various adjacent topics, and whatever else interests me, Stilgherrian, published every Friday afternoon (nearly).
If I’ve missed anything, or if there’s any specific items you’d like me to follow, please let me know.
If you find this newsletter useful, please consider throwing a tip into the tip jar.
This is not a cyber security newsletter. For that that I recommend Risky Biz News and Cyber Daily, among others.