RESEARCHERS QUESTION ANTHROPIC CLAIM THAT AI-ASSISTED ATTACK WAS 90% AUTONOMOUS

Summary from the TLDR InfoSec newsletter

Anthropic reported a major AI-driven cyber espionage attempt using Claude AI, claiming it automated 90% of hacking steps for suspected Chinese attackers. However, outside experts challenge this view, saying such full automation is exaggerated, as most attacks were not successful, and the AI often fabricated results.