Anthropic to sign the EU Code of Practice

After review, Anthropic intends to sign the European Union's General-Purpose AI Code of Practice. We believe the Code advances the principles of transparency, safety and accountability—values that have long been championed by Anthropic for frontier AI development. If thoughtfully implemented, the EU AI Act and Code will enable Europe to harness the most significant technology of our time to power innovation and competitiveness.

A recent analysis found that AI has the potential to add more than a trillion euros per year to the EU economy by the mid-2030s. The Code, working alongside Europe's AI Continent Action Plan, demonstrates how flexible safety standards can both preserve innovation and enable broader AI deployment. This approach highlights the opportunities and imperatives required for Europe to remain competitive in this transformational technology. With transparent risk assessment processes in place, we can accelerate work to address Europe's most pressing challenges: advancing scientific research, improving public services, and enhancing industrial competitiveness.

We're already seeing signs of what's possible, from Novo Nordisk accelerating breakthrough drug discovery, to Legora transforming legal work, to the European Parliament expanding access to decades of archives to citizens. Ensuring these benefits materialize with minimal downside requires public visibility into AI safety and transparency practices while preserving private sector agility to deliver AI's transformative potential.

Building on our commitment to transparency

As outlined previously, Anthropic believes the frontier AI industry needs robust transparency frameworks that hold companies accountable for documenting how they identify, assess, and mitigate risks. The EU Code establishes this baseline through mandatory Safety and Security Frameworks that build upon Anthropic’s own Responsible Scaling Policy and will describe important processes for assessing and mitigating systemic risks. This includes assessment of catastrophic risks—particularly those from Chemical, Biological, Radiological, and Nuclear (CBRN) weapons.

Maintaining flexibility

AI moves fast and changes constantly, which means the best policies are those that can be flexible and adapt alongside the technology.

Over the nearly two years since we first published our Responsible Scaling Policy, we've refined it several times based on practical insights from implementation. For example, our most recent update clarified which actors are in-scope for the ASL-3 Security Standard; this determination was based on a deeper understanding of the relevant threat models and model capabilities.

As an industry, we're still developing best practices for assessing the systemic risks identified in the Code. Different risks require different methodologies. Third-party organizations like the Frontier Model Forum play a critical role, establishing common safety practices and evaluation standards that evolve with the technology. These groups bridge industry and government, translating technical insights into actionable policy.

We're committed to working with the EU AI Office and safety organizations to ensure the Code remains both robust and responsive to emerging technologies. This collaborative approach—combining regulatory frameworks with flexibility—will be essential for Europe to harness AI's benefits while competing effectively on the global stage.