EU Releases Code of Conduct for General-Purpose AI Amidst Debate

The European Commission has finally published the code of conduct for general-purpose artificial intelligences (AI) on Thursday, July 10. This text, intended for large text or image generation models that serve as the basis for business uses or assistants like ChatGPT (OpenAI), Gemini (Google), or Le Chat (Mistral), was initially expected on May 2. Despite this delay, Brussels has announced that it will maintain the date of August 2 for the entry into force of this part of the AI Act’s obligations. However, this European AI regulation is still the subject of intense debates and pressures, particularly from American giants in the sector and European companies, some of whom are calling for a “pause” in its implementation. While displaying its firmness, the Community executive does not close the door to certain adjustments.

In concrete terms, the code of good practice will oblige its signatories to provide technical documentation to all companies and administrations that want to integrate their models into their products or activities. Manufacturers will also have to put in place a “copyright policy” committing them not to train their AI with content found on “pirate” sites or owned by rights holders who have refused this use (“opt-out”), and also to take measures to prevent content produced by the models from plagiarizing protected images, sounds, or texts.

The remainder of this article is available for subscribers.

---

Enjoyed this post by Thibault Helle? Subscribe for more insights and updates straight from the source.