OpenAI has released a blueprint outlining proposals for governments and industry to strengthen cybersecurity in the context of advancing artificial intelligence, according to a statement from the company.

The document builds on OpenAI's earlier cybersecurity action plan and frontier safety blueprint. The company said its own experience with misalignment and security incidents involving its models informed the proposals.

The blueprint is organized around three areas. The first calls for strengthening safety and security for advanced AI by establishing clearer responsibilities for securing frontier models, detecting dangerous behavior, and maintaining shutdown controls. The second focuses on modernizing cybersecurity by equipping defenders with advanced tools, training, and technical support, while addressing longstanding weaknesses and investing in essential services. The third proposes building more resilient public institutions through stronger government evaluations, independent technical assessments, consistent incident reporting requirements, and improved coordinated response and recovery.

OpenAI stated that AI can assist defenders in finding and fixing vulnerabilities, detecting intrusions, and responding to attacks, but also acknowledged that the same capabilities can be used by malicious actors and can create new risks when AI systems operate beyond their intended boundaries.

The company described the current moment as a "defender's window," characterizing it as a limited opportunity to place advanced AI into the hands of defenders before advanced offensive cyber capabilities become widely available.

OpenAI said AI laboratories have a responsibility to address risks created by their systems, and called on governments to develop the expertise, resources, and authority needed to protect the public as AI capabilities advance.