EU Issues First Guidance on AI Safety Standards for Developers

EU Issues First Guidance on AI Safety Standards for Developers

The European Union published a General-Purpose AI Code of Practice on July 10, 2025, providing the first concrete guidance on how AI developers should comply with the EU's AI Act, which took effect in June 2024.

The code addresses a critical gap in the landmark regulation: while the AI Act requires providers of general-purpose AI systems to "assess and mitigate possible systemic risks," it does not specify how. The ambiguity has left developers uncertain about compliance and regulators unable to verify whether safety standards are being met.

The code applies to AI systems trained using 10^25 FLOPs or more—those deemed to pose systemic risk. Beyond safety standards, the AI Act also imposes transparency requirements and copyright protections on general-purpose AI providers.

Why this matters for finance leaders: CFOs evaluating AI tools for financial forecasting, credit analysis, or other high-risk applications must now track compliance with both the AI Act and this new code of practice. Non-compliance carries regulatory risk in EU markets.

The guidance represents the first major implementation test of the world's most comprehensive AI regulation. Watch for how major AI providers respond and whether the code becomes a de facto global standard.