Anthropic announced Thursday that Chinese state-backed hackers used its AI model Claude to automate about 30 attacks against businesses and governments during the September campaign. wall street journal.
Anthropic says that up to 80% to 90% of the attacks are automated by AI, which is higher than any previous hack. Jacob Klein, head of threat intelligence at Anthropic, said the incident happened “literally at the click of a button and with minimal human intervention.” journal. He added, “The humans just got caught up in a few serious conundrums: ‘Yes, continue,’ ‘Don’t continue,’ ‘Thank you for this information,’ and ‘Oh, that doesn’t seem right, Claude, are you sure?'”
AI-powered hacking is becoming more and more common, as are modern strategies that use AI to bring together the various tasks required for a successful attack. Google has discovered that Russian hackers are using large-scale language models to generate commands for malware, according to a company report released on November 5.
The U.S. government has long warned that China is using AI to steal the data of American citizens and businesses, a charge China has denied. Antropic said: journal He said he believes the hackers were backed by the Chinese government. In this campaign, the hackers stole sensitive data from four victims, but like previous hacks, Anthropic did not reveal the names of the targets, whether successful or not. The company said the U.S. government was not a successful target.
