Forum

A Rulebook and a Collapsing Building: Two Signals From July

Upward view of a large textured building corner rising into a blue sky with soft clouds

Two announcements from the past two weeks belong in the same conversation, even though one came from Brussels and the other from a streaming company’s earnings call.

On 10 July, the European Commission published the final version of its General-Purpose AI Code of Practice, a voluntary guide for companies building large models to meet obligations under the EU AI Act that begin to apply on 2 August. A week later, on 17 July, Netflix co-CEO Ted Sarandos told investors that the Argentine series El Eternauta contains what he called the very first generative AI final footage to appear on screen in one of its productions.

One is a rulebook for model makers. The other is a sign of how quickly those models are moving into finished creative work.

What the code asks of model makers

The code has three chapters: transparency, copyright and safety and security. The copyright chapter is the one creative professionals should read. Signatories commit to keeping a copyright policy, respecting machine-readable signals from rights holders who do not want their work used for training, and not circumventing technical measures that protect content.

None of this is new law; the obligations come from the AI Act itself. The code is a way for companies to show they comply, and signing is optional. That optionality was tested quickly. On 18 July, Meta said it would not sign, citing legal uncertainty and measures it argued go beyond the Act. Other developers, including OpenAI and Mistral, have indicated they intend to sign.

For artists and designers, the practical meaning is modest but real. Machine-readable opt-outs gain weight in Europe, and model providers will have to describe their training content in more detail than before. Enforcement by the Commission’s AI Office begins later, but the direction is set.

What Netflix actually did

The El Eternauta example is specific. According to Sarandos, Netflix’s in-house production innovation team used generative AI to create a sequence showing a building collapsing in Buenos Aires. He said the shot was completed about ten times faster than it could have been with traditional visual effects workflows, and at lower cost, and framed the technology as a way to help creators make films better, not only cheaper.

The scene is brief. Its significance lies in what it represents: a major studio describing generative output as final footage in a successful series, and presenting that as a selling point to investors rather than something to downplay.

Where these stories meet

Put the two announcements side by side and a gap appears. European rules focus on the companies that build general-purpose models. Netflix’s example concerns the companies and creative teams that use them. Responsibility for what ends up on screen, in a campaign or on a gallery wall sits largely with those users.

That is where designers, studios and visual effects artists come in. The questions they face are not covered by the code:

  • Which model produced the footage or image, and under what terms was it trained and licensed?
  • Who is credited for a shot that combines generated material with human direction, compositing and grading?
  • What happens to the artists whose work used to fill those days of production?
  • How, if at all, should audiences be told?

A view from the creative side

It would be easy to read the Netflix story as a threat and the EU code as a shield. Neither reading is quite right. The code does little for individual artists in the short term, and a single collapsing building does not replace a visual effects department.

What both stories do is normalize. Generative tools are moving from experiments into regulated, budgeted, credited production. In that world, the professionals who stand out will be those who can explain their pipelines as clearly as they can execute them: what was generated, what was crafted, and why the combination serves the story.

For award entrants

The 3rd International AI Award 2025 announces its winners at the end of this month. Whatever the results, the categories on the Rules page, from AI-driven film and animation to interactive experiences, reflect the same reality Netflix described: AI is now part of finished work across disciplines, and transparency about its role is part of the craft.