Artificial IntelligenceHigh Priority (9/10)

Anthropic's New AI Model 'Mythos' Poses Major Cybersecurity Risks

Anthropic is privately warning government officials that its upcoming 'Mythos' model could enable large-scale cyberattacks with unprecedented sophistication, potentially leading to attacks this year.

Key Points

  • Anthropic's upcoming 'Mythos' model could enable large-scale cyberattacks
  • New AI models allow agents to work autonomously with unprecedented sophistication
  • Government officials warned that attacks could occur this year
  • Employees increasingly using AI agents at home creates additional vulnerability

Full Details

Top AI and government officials tell Axios that Anthropic, OpenAI, and other tech giants will soon release new models with alarming capabilities for conducting sophisticated cyberattacks at scale. Anthropic is privately warning top government officials that its not-yet-released model, currently branded 'Mythos,' makes large-scale cyberattacks much more likely in 2026. The model allows AI agents to work autonomously with remarkable sophistication and precision to penetrate corporate, government, and municipal systems. Unlike previous AI models, these new systems can think, act, reason, and improvise independently without rest or limitation. Simultaneously, systems are more vulnerable because employees are increasingly using Claude, Copilot, and other agentic models at home, creating their own AI agents. One source briefed on the coming models says a large-scale attack could hit this year.

Why It Matters

The convergence of highly capable AI systems and widespread adoption of autonomous AI agents creates unprecedented cybersecurity challenges that require immediate policy attention and defensive measures.

Sourceaxios.com

Get stories like this delivered daily

AI-curated news, personalized to your interests. Zero noise.

Start 7-Day Free Trial →

More in Artificial Intelligence