Anthropic and OpenAI Unveil Cyber Models

The race for AI supremacy has shifted toward defensive and offensive cybersecurity capabilities as labs release increasingly powerful models.

Anthropic's Mythos has already demonstrated the ability to complete complex multi-step infiltration challenges, prompting the U.S. Treasury to seek access for vulnerability testing.

OpenAI's response, GPT-5.4-Cyber, is designed to assist in finding software bugs but remains restricted due to potential misuse risks.

Industry leaders like Jamie Dimon have warned that these models could uncover significant new vulnerabilities, necessitating a rapid regulatory response.