The Weekly Cybers #89
Deloitte’s corrected report needed more corrections, discourse continued as Australia’s social media age restrictions loom, and much more.
17 October 2025
Welcome
Australia’s social media age restrictions get closer but the discourse is far from dead. We can expect plenty more of it before 10 December, and indeed beyond.
As usual there’s plenty of AI-related stories this week, including news that the corrected version of Deloitte’s AI-riddled report for government was still riddled with AI.
And there’s — hey you can read it all for yourself. Enjoy.
Flurry of action ahead of social media age rules
It’s 54 days until the start of Australia’s social media age restrictions. The government has released the first adverts, and resources for families more generally. But there’s still plenty of open questions.
Remember, as the eSafety Commissioner says, “It’s not a ban, it’s a delay to having accounts”.
Which means, as the Guardian has demonstrated, under-16s could well be able to see bad things without logging in.
Google has launched a final push to avoid the restrictions, telling a Senate committee the law would be “extremely difficult” to enforce.
Instagram will be restricting teens’ accounts to content similar to the US PG-13 movie rating, which in turn is similar to to Australia’s M rating.
Meanwhile, advocates say the ban — sorry, the delay — will further isolate regional and rural youth and LGBTIQA+ teens.
Finally, stepping back a bit, The Conversation has a discussion of whether the internet is entering its Victorian era of increasing social conservatism.
The next few weeks are certain to see more special pleading on behalf of the platforms, more nations announcing their intention to follow Australia’s lead, and more discussion of whether the restrictions are really such a good idea and whether than can even work as intended. Watch this space.
Oh no, FoI! One email every five minutes!
I find it very hard not to laugh at the Guardian report which revealed an unnamed eSafety staffer worried that one email every five minutes might “jam something”.
The eSafety office certainly received a surge of freedom of information (FoI) requests, some 600 in a relatively short period of time, via a pre-filled form at Right to Know — that’s a website designed to help people file their own FoI requests.
This specific form was set up for people wanting to know what information eSafety held about their X account.
Attorney-General Michelle Rowland says this is an example of “why the current system is broken”, and it’s one of the arguments she puts forward to support new legislation to restrict FoI and charge people to file requests.
The opposition leader says they’ll vote against the bill, but the A-G is adamant the legislation will go ahead. I guess we’ll soon see.
One might argue, however, that the proper way to deal with an increase in FoI requests would be to hire more staff to handle this legal requirement, or simply to publish more information to begin with.
Meanwhile, Services Australia says its FoI disclosure log will finally be online by the end of the year.
“Corrected” Deloitte report still has AI errors
As we wrote last week, global consulting firm Deloitte was caught using AI to write a $440,000 government report, and had to issued a corrected version.
Well, as the Financial Review reported ($), “The corrected version ... contains new citation errors and other flaws in referencing the academic work it relies on”.
Correction: Despite last week’s original headline, Deloitte is refunding only part of their $440,000 fee, specifically $97,587.11.
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.
Also in the news
- The Australian Signals Directorate (ASD) has released the Annual Cyber Threat Report 2024-2025. This is always a solid overview, although your somewhat jaded writer notes that it always seems to be a tad depressing.
- The ASD is “probably one of the busiest arms of government”, according to director-general Abigail Bradshaw. The $10 billion REDSPICE program has seen a “huge uplift in ASD capability that includes almost doubling the size of the workforce,” she said.
- As part of that being-busy, ASD has published updated cyber advice for network defence.
- The Digital Transformation Agency (DTA) has published new guidance for its Australian Government colleagues on the “safe and responsible use of public generative AI tools”.
- The personal data of up to six million Qantas customers has been leaked into the dark web as cyber extortionists made good on their demand. But don’t worry, Qantas got a court injunction to prevent anyone downloading it. Alas, that included security researchers who might otherwise have been able to warn people about the breached data.
- South Australia has banned AI deepfakes in political advertising.
- Australia will be cracking down on cryptocurrency ATMs and has granted banks access to more information about customers. Other anti-fraud plans were outlined by home affairs minister Tony Burke in his address to the National Press Club on Thursday.
- “The Australian government is set to launch its long-awaited National Driver Licence Facial Recognition Solution by the end of the year, nearly a decade after the system was first proposed,” reports iTnews. This was a system once dubbed “The Capability”, but the enabling legislation was found to be so privacy-invasive that the Parliamentary Joint Committee on Intelligence and Security (PJCIS) demanded the whole thing be scrapped, new legislation drawn up, and a new start be made. This is the result.
- Shadow communications minister Melissa McIntosh has serious questions about the ACMA’s ability to do the job.
- The National Archives has identified four “areas of interest” for using AI with their 10 petabytes of data.
- Two years after mobile phone were banned in most Australian schools, as one principal put it, “the gains are real, modest and worth it”.
Elsewhere
- As WIRED reporter Matt Burgess put it, “Oh good, ChatGPT is getting ‘erotica for verified adults’ later this year”.
- The Carnegie Endowment has some analysis of California’s new AI law, which is focused on increased transparency.
- From The Conversation, “AI systems and humans ‘see’ the world differently — and that’s why AI images look so garish”.
- “Wikipedia’s data shows that AI is siphoning traffic away from the site, which is a danger to its sustainability,” posted 404 Media’s Emanuel Maiberg. “Ironically Wikipedia is more important than ever to users who want reliable information instead of slop, and to AI companies that need it for training data.”
WHY NOT LISTEN TO OUR PODCAST ABOUT MUSIC? My good friend Snarky Platypus and I have posted a fresh episode of Another Untitled Music Podcast. Look for it in your podcast app.
Inquiries of note
Nothing new this week.
What’s next?
Parliament is currently on break. Both houses return on Monday 27 October, which is 10 days away, for two weeks of sitting.
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.