Tapping the powers of Mythos-like models still requires human intervention

AXIOS

Anthropic and OpenAI’s cyber-capable AI models may still require significant human expertise to operate effectively, according to new findings from users testing the systems in real-world environments.

Why it matters: The new phase of AI-powered cybersecurity may depend less on fully autonomous hacking and more on how effectively humans can direct, validate and operationalize increasingly powerful systems.

The big picture: When Anthropic unveiled Mythos Preview to the world, it warned that the model was so powerful that it found tens of thousands of bugs spanning nearly every operating system.

  • Third-party testing suggests that OpenAI’s GPT-5.5-Cyber is roughly as powerful as Mythos at finding bugs and writing exploits, though Mythos may have an edge in some more elaborate attack scenarios.
  • Major companies and governments around the world have been clamoring to get their hands on these models to understand what they’ll be up against once similar capabilities fall into the hands of attackers.

Discuss

Here is where members can discuss, give feedback, and present their ideas within the “Tapping the powers of Mythos-like models still requires human intervention” post. OnAir membership is required to participate.

The lead moderator for the discussions is Zeinab Shariff. We enforce civil, honest, and respectful discourse across our network of hubs. For more information on commenting and giving feedback, see our Community Guidelines.

This is an open discussion on this news piece.

Home Forums Open Discussion

Viewing 1 post (of 1 total)
Viewing 1 post (of 1 total)
  • You must be logged in to reply to this topic.
Skip to toolbar