The European Commission has launched its project to develop the first-ever code of conduct for general AI, closely linked to the recently adopted EU AI law.
The aim of the code is to set clear ground rules for AI models like ChatGPT and Google Gemini, particularly on the subject of things like transparency, copyright and how you can manage the risks that these powerful systems pose.
At a recent online plenary session, nearly 1,000 experts from academia, industry and civil society gathered to assist shape what this code will appear like.
The process can be led by a gaggle of 13 international experts, including Yoshua Bengio, one among the “godfathers” of AI, who will lead the group with a give attention to technical risks. Bengio won the Turing Award, which is effectively the Nobel Prize in computer science, so his opinions carry deserved weight.
Bengio's pessimistic views on the catastrophic threat that powerful AI poses to humanity foreshadows the direction the team he leads will take.
These working groups meet commonly to draft the code. The final version is predicted to be available in April 2025. Once finalized, the code could have a significant impact on any company that desires to deploy its AI products within the EU.
The I HAVE Act sets out a strict regulatory framework for AI providers, however the Code of Conduct can be the sensible guide that firms must follow. The code addresses issues similar to increasing the transparency of AI systems, ensuring they comply with copyright laws, and establishing measures to deal with the risks related to AI.
The teams drafting the code will need to contemplate how you can develop AI responsibly and safely without stifling innovation, something the EU is already criticized for. The latest AI models and capabilities from Meta, Apple and OpenAI aren’t fully deployed within the EU on account of already strict GDPR data protection laws.
The effects are enormous. If properly implemented, this code could set global standards for AI safety and ethics and provides the EU a leadership role in AI regulation. However, if the code is just too restrictive or unclear, it could slow AI development in Europe and drive out innovators elsewhere.
While the EU would undoubtedly welcome the worldwide rollout of its code, that is unlikely as China and the US look like more pro-development than risk-averse. The veto of California's AI safety law SB 1047 is a very good example of the several approaches to AI regulation.
AGI is unlikely to emerge from the EU tech industry, but it is usually less likely that the EU is the start line for a possible AI-powered disaster.
