The Group of Chief Scientific Advisors (GCSA) and the European Group on Ethics in Science and New Technologies (EGE) have published a joint statement on artificial intelligence. Its title is “The next phase of artificial intelligence: governing for safety, security and human agency”.
Artificial intelligence is advancing quickly, and governance must address risks from individual applications as well as systemic risks. The Group of Chief Scientific Advisors and the European Group on Ethics in Science and New Technologies argue that regulation must keep pace so society’s understanding and governance of emerging capabilities and risks continue to improve.
Europe already has a substantial legal and institutional framework, notably the EU AI Act. The priority is its effective application, implementation, monitoring and enforcement, with safety and security remaining prerequisites for AI development and deployment.
The Group of Chief Scientific Advisors and the European Group on Ethics in Science and New Technologies set out four principles:
- No compromising of current legislation: safety and security must remain prerequisites for development and deployment, with evidence requirements proportionate to capability and potential consequences.
- Not every AI system should be developed and deployed at the same speed: lower-risk applications may progress more rapidly, while high-risk, highly autonomous or difficult-to-reverse deployments need stronger testing, human oversight and effective safeguards.
- Europe must retain the capacity to independently evaluate and govern the AI systems it depends on, independently of the companies that develop them, especially systems developed outside Europe. This means strengthening scientific and technical expertise, shared evaluation infrastructure and technological sovereignty, while reducing strategic dependencies.
- Build an AI safety, security and learning ecosystem, based on continuous monitoring, independent evaluations, incident reporting, learning from failures and near misses, and international cooperation.
A proportionate, risk-based approach allows Europe to keep benefiting from advances in AI while safeguarding security, fundamental rights, democratic accountability and human agency.
A statement is a shorter format than a Scientific Opinion. Rather than building on a SAPEA Evidence Review Report, it gives the two groups’ collective view on a fast-moving question.
Read the joint statement on governing AI for safety, security and human agency.
