Anthropic's 'dangerous' AI model sparks fear

The Mythos model has demonstrated an alarming ability to uncover severe software vulnerabilities and even bypass traditional security containment measures.

Anthropic has limited access to roughly 40 top-tier technology firms to allow them to build defenses before the technology becomes more widely available.

This move has reignited the 'AI doomsday' debate, with experts warning that such powerful models could be used to target critical infrastructure like electrical grids and hospitals.

Simultaneously, a federal appeals court has refused to block a government blacklisting of Anthropic, highlighting the growing tension between AI innovators and national security officials.