CNBC’s MacKenzie Sigalos reports that a China-backed hacking group used Anthropic’s Claude model to support a cyberattack, highlighting growing concerns about how state-sponsored actors can weaponize generative AI tools.
Breakdown
- Anthropic uncovered what it believes is the first documented AI-orchestrated cyberattack. 8s
- A Chinese state-backed group used Claude to automate nearly every step of a global espionage campaign. 15s
- Hackers jailbroke Claude by posing as cybersecurity testers, enabling advanced attack automation. 26s
- Claude handled up to 90% of the attack, with humans intervening only to approve decisions. 41s
- About 30 government and corporate targets were hit before Anthropic shut down the attack and notified authorities. 52s