1 May 2026

Welcome

Remember how last week I wrote about how digital technology changes power relationships? Well we’ve got another great example this week: the struggle over who chooses what news and other content we see.

A group of experts says there’s as “urgent need for transparency in algorithmic curation”, what with the potential for misinformation.

That’s even more important when you consider that the major online platforms — search and social media — are increasingly the first encounter people have with stories. Their algorithms choose the priority, now AI summarises the content (somehow), and the platform scoops up the advertising money — leaving far fewer click-throughs to the news mastheads who did the original journalism.

Which is why the government is having another go at a news bargaining code.

Also this week, a surfeit of goblins. Read on...

Platforms to pay for news content, round two

The Australian government is having another go at “encouraging” the big online platforms to pay news outlets for news content, this time calling it the News Bargaining Incentive.

“Under the Incentive, digital platforms operating significant social media or search services are encouraged to do commercial deals with eligible news publishers as the preferred model, with generous offsets provided to reduce their liabilities,” says the PM’s press release.

“Platforms who elect not to do commercial deals with news publishers will need to pay a charge as a proportion of their revenue, with any charges collected to be distributed back to the news media sector.”

That tax is proposed to be 2.25% of a platform’s Australian revenue.

To put this into context, four of the tech giants — Meta, Alphabet, Microsoft, and Amazon — together earned more than US$100b in advertising revenue globally in the first three months of 2026.

As you might expect, big tech has complained, while news outlets generally welcomed the idea.

Meta called it a “government-mandated transfer of wealth” with “no connection to the value exchanged”. It’s worth remembering that last time, under the News Media Bargaining Code, Meta simply avoided the code by deprioritising news.

Google has asked why Microsoft is exempt, for example. Presumably it’s because the latter’s social media and search revenue in Australia is less than $250 million, or its user numbers are below the user thresholds — five million average active monthly users for a “significant media service” and 10 million for a “significant search service”.

A Trump administration official has said: “President Trump is committed to defending America’s leading technology sector from digital services taxes and other forms of foreign extortion.”

An opinion piece in The Conversation argues that this latest plan is a high-risk move.

Quite a few questions still need to be answered: How do you define journalism in an age of influencers, social media and chatbots? Who qualifies for funding? Are the current eligibility criteria fit for purpose? And will this include influencers such as Konrad Benjamin, aka Punter’s Politics, who has large audiences for his explainer reporting?

The Local and Independent News Association (LINA), which represents over 170 independent newsrooms across Australia, joined with the Community Broadcasting Association (CBAA) to welcome the legislation, but also to question its effectiveness.

“Commercially focused eligibility requirements create barriers to participation for small publishers and volunteer-based community media, further entrenching the status quo in a media landscape that is already one of the most concentrated in the world,” they wrote.

Treasury has published the exposure draft legislation and explanatory material. The consultation period is short. Submissions close 18 May.

And anyway, how do they choose what news we see?

A bunch of experts reckons it’s time for the big platforms to tell us how AI chooses what we see.

“There is an urgent need for transparency in algorithmic curation and mandatory labelling of AI-generated content,” they write.

LATEST PODCAST: If you read this newsletter, and it appears that you do, then may I suggest listening to The 9pm Cornucopia of Tech Policy Pleasures with Johanna Weaver and Zoe Jay Hawkins from the Tech Policy Design Institute in Canberra? In this episode we talk about how Australia as a middle power can participate in global tech policy. We chat about AI slop, the battle between Anthropic and the Pentagon, the digital duty of care, and of course the social media age restrictions. Just look for The 9pm Edict in your podcast app.

Also in the news

• Calls are growing to ban Palantir from Australia. While Palantir insists it’s just a software company, it’s recently published a fascist-adjacent political manifesto, and it certainly has a PR problem these days.

• The Australian Public Service Commission (APSC) has published a set of expectations for government agencies using AI in recruitment, with a deadline for implementation of 1 July. And if you’re applying for a job, Principles for candidate use of AI in recruitment (PDF).

• ABC News has an explainer on how to handle deepfakes in schools. Meanwhile the Tasmanian private school at the heart of this story denies a cover-up.

• eSafety commissioner Julie Inman Grant has drawn attention to the online abuse of women leaders, noting that elected officials have security protections but regulators do not. In the first 24 hours after Elon Musk called her a “censorship commissar”, some 75,000 posts had been directed at her, “80% of which were toxic, harmful, and plausible death threats,” reports The Mandarin.

• “Cyber attacks on medical devices pose ‘significant’ impact on real-life patient care,” reports Cyber Daily. Yes, medical devices can be hacked. It seems little has improved since the late Barnaby Jack wirelessly hacked an insulin pump to deliver a lethal dose some 15 years ago.

• Unsurprisingly, medical AI is moving faster than safety checks, according to new research from Flinders University.

• From Pivot to AI, an explanation of “tokenmaxxing”, a concept which is definitely not self-serving for Nvidia nor stupid for everyone else. Also, how do you feel about the suffix “-maxxing”?

• Letting AI write our emails might actually create more work, argues QUT’s Professor Daniel Angus.

PLEASE SUPPORT THIS NEWSLETTER: 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

• Pushback against online age verification is increasing. A group of security and privacy academics across 29 countries has said age checks might be ineffective, and can even “diminish safety online by exposing users to malware and scams when they resort to alternative services that do not implement verification”.

• The European Commission has recommended EU member states adopt its age verification app even though, as we wrote last week, it can be bypassed very easily.

• Meta has been found to have breached EU law by failing to keep under-13s off Facebook and Instagram.

• South Africa has withdrawn its AI policy after it was found to have been written by AI.

• The Internet Corporation for Assigned Names and Numbers (ICANN) has opened applications for new generic top-level domains (gTLDs), the part of a web address after the last dot. Domains are now available in 27 different scripts, representing hundreds of languages such as Arabic, Chinese, Devanagari, and Thai. Applications close 12 August.

• The Netherlands has released an open-source code platform for government, a sovereign European alternative to US-based GitHub.

• This week's ASPI’s Cyber and Tech Digest has a good piece on China’s DeepSeek V4 AI, as well as plenty more of interest.

• From Scimex, “A Chinese study of Americans’ use of artificial intelligence (AI) found that wealthier and more educated people are more aware of AI, more familiar with it, and more likely to use it than those from a more modest background.”

• Elon Musk and OpenAI’s Sam Altman are involved I a bog far court battle. I haven’t been following it, so here’s an update from the New York Times (gift link), as well as a link to their ongoing live coverage. Also from the NYTimes, what it’s really about (another gift link).

• OpenAI has published a blog post about goblins, or rather, “[why] our models began developing a strange habit: they increasingly mentioned goblins, gremlins, and other creatures in their metaphors”.

• A Peter Thiel-funded startup called Objection AI says it will use an “AI jury” to “subject the media’s claims to systematic investigation and judgment”. Or as Oligarch Watch puts it, to discredit journalism with AI by flooding the zone with plausible-looking counterarguments.

• Information scientist Carl T Bergstrom has posted some excellent examples which illustrate how asking AI models about things you know well quickly exposes the garbage.

• Microsoft has reorganised its Windows team to win back trust for Windows 11, an operating system that hasn’t exactly been well-received.

Inquiries of note

Apart from the News Bargaining Incentive consultation described earlier...

• Treasury is looking at how the Data Standards Body (DSB) makes regulatory data standards for the Consumer Data Right (CDR) and Digital ID. Exciting stuff. Submissions close 29 May.

What’s next?

Parliament is on break until Tuesday 12 May, when it’s Budget Night. That’s just 11 days away. But I’ll be watching out for public hearings for the various inquiries under way. And for relevant Budget leaks.

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.