Anthropic to sign EU’s General-Purpose AI Code of Practice
Anthropic has confirmed its intention to sign the EU’s General-Purpose AI Code of Practice, aligning with European efforts to promote transparency, safety, and accountability in frontier AI development.
Anthropic announced on 21 July 2025 that it intends to sign the European Union’s General-Purpose AI Code of Practice. The company stated that the Code aligns with its long-standing values of transparency, safety, and accountability in AI development. The move comes amid broader EU efforts to operationalise the AI Act and foster responsible innovation through frameworks like the AI Continent Action Plan.
The EU Code sets out mandatory Safety and Security Frameworks and emphasises public visibility into AI risk management processes. Anthropic noted that this aligns with its Responsible Scaling Policy, which includes assessment of systemic risks, such as those arising from chemical, biological, radiological, and nuclear threats.
The company also emphasised the importance of policy adaptability in response to rapidly evolving AI technologies. Anthropic’s updates to its own internal policies, such as refining threat model scopes under its ASL-3 Security Standard, reflect this need for ongoing revision based on real-world insights.
Anthropic underlined the role of collaborative institutions like the Frontier Model Forum in developing shared safety standards. The company expressed its commitment to continued work with the EU AI Office and other safety organisations to ensure the Code remains effective and adaptive.
By endorsing the Code, Anthropic joins a growing number of AI developers supporting Europe’s efforts to balance innovation with safety in the deployment of advanced AI systems.
