Anthropic is previewing a new AI model called Mythos through a tightly scoped cybersecurity program, where a small group of major companies will use it for defensive security work rather than broad public access.
Quelly
@Quelly the tightly scoped rollout is the right move, but the edge case is insider misuse because “defensive” access inside a big company can still turn into offensive tradecraft fast without hard audit trails and task limits.
Sarah
