The initiative stresses tech sovereignty and addresses gaps in access to European evaluators and opaque third-party criteria, including for projects like Anthropic.

The plan is accompanied by materials such as a press release and a factsheet outlining its scope and implementation steps.

Industry voices urge stronger actions than guidance, highlighting debates over pre-market evaluations and regulatory reach amid fast-evolving AI-enabled threats.

Officials note that while cost is a concern, not developing leading AI models would be more costly in the long run.

Tensions between security and civil liberties are acknowledged, with concerns over access, oversight, data handling, and proportionality in AI-enabled security tools.

ENISA will foster public-private and open-source partnerships, publish guidance and best practices, and launch campaigns to secure critical open-source software.

The EU is planning a dedicated evaluation capacity for AI with a cybersecurity focus, building European evaluation infrastructure and a blueprint for structured access programs to let EU bodies access advanced AI while mitigating cyber risks.

Together with the Joint Research Centre, ENISA will create a secure platform to test AI for cybersecurity, including simulated environments, to promote safe usage across finance, energy, health, transport, and public administration.

Regulatory changes are imminent, and industry observers urge preparing now for tomorrow’s shifts.

The episode where the US briefly blocked access to two advanced AI models for vulnerability detection has ended with export controls lifted after about two weeks in early July.

Executive Vice-President Henna Virkkunen unveiled the plan during a plenary session in Strasbourg.

MLex is referenced as a source of ongoing news, analysis, and predictions on AI, cybersecurity, and related regulations.