ChatGRT and the EU act on artificial intelligence

General Artificial Intelligence (AGI) - also known as Generative AI - has been prominent in the news following the release of large language models such as ChatGPT from OpenAI and Bard from Google.

In the area of copyright, Getty Images recently announced that they are suing Stability AI, the creator of AI-based image generator Stable Diffusion, for copyright infringement.

With the draft Artificial Intelligence Act (AI Act) currently being considered by the European Parliament, recent speculation suggests that the European Parliament is considering including AGI in the "high risk" category of AI systems.

It is possible that AGI would be considered a transparent category, but its end use in products could be high risk. In other words, the decision will be based on the end use rather than the model itself.

AGI refers to a category of algorithms that generate new results based on the input data on which they have been trained.

Title IA in the new draft of the II

Act.

The European Council has added a new heading IA in its draft AI Act. Although it contains only three paragraphs, a new heading IA has been added to take into account situations where AI systems may be used for many different purposes and where AGI technology is integrated into another system that may itself become high risk.

The European Council text specifies in Article 4b(1) that certain requirements for high risk AI systems will also apply to AGI.

The draft AI Act defines a "generic target AI system" as a system that, according to the vendor's intentions, performs generic functions such as image and speech recognition, audio and video generation, pattern detection, question answering, translation and others. This definition applies regardless of how the system is placed on the market or commissioned, including open source software.

The AGI system can also be used in a multitude of contexts and integrated into many other AI systems.

AGI system providers will require a significant amount of compliance documentation, including the compilation of technical documentation in accordance with the provisions of Section 11 of the AI Act.

This technical documentation must demonstrate that the high-risk AGI system complies with the requirements set out in the AI Act. In addition, the documentation shall provide the national authorities of competence and notified bodies with all the necessary information in a clear and complete form to assess the compliance of the AI system with these requirements.

This documentation should be available to the national authorities of competence for 10 years after the general purpose system has been placed on the market or put into operation in the EU.

With regard to exemptions, Article 4b of the AI Act states that it does not apply if the supplier has explicitly excluded all high-risk uses of AI in the instructions or information accompanying the general purpose AI system.

In conclusion, we await the final text of the AI Act to see if AGI will be classified in the same category as "high risk" AI systems.

Recent reports say lawmakers may provide for AGI in the transparency category, but its end use in products could be high risk.

The European Parliament is expected to finalize its position on the AI Act in the coming weeks. The intensive nature of the negotiations and the continuous reports of progress and compromises indicate the willingness of European Union lawmakers to meet the planned deadlines for the adoption of the law by Parliament and its transmission to the Commission.