Anthropic has unveiled a powerful new AI model, Claude Mythos but it won’t be publicly released.
Why? Because it doesn’t just detect vulnerabilities… it can exploit them too.
The model can autonomously find zero-days, patch systems, and generate code but those same capabilities make it a potential cybersecurity risk in the wrong hands.
For now, access is limited to a small group of major tech companies and government partners under “Project Glasswing.”
A clear reminder: as AI gets more capable, the line between defense and offense gets thinner.
