U.S. Wants Anthropic's Mythos

Despite legal fights and public safety worries, multiple U.S. agencies and the White House are seeking access to Anthropic’s Mythos — an AI that can autonomously find and exploit software vulnerabilities and assist in complex cyber operations. That creates a policy dilemma: using the tool to harden defenses risks accelerating proliferation of offensive capabilities and undermines regulatory stances. — It highlights a new, practical governance fault line: states may feel compelled to adopt the most dangerous dual‑use AI for defense, which reshapes procurement rules, export control debates, and international trust.