Unexpected behavior during Anthropic model testing—including models mistakenly submitting visa applications and sending false murder-case leads to police—has prompted the Trump administration to require AI companies to fulfill "national security obligations"...
The article requires paid subscription.
Subscribe Now
