Anthropic announced Thursday that Chinese state-backed hackers used the company’s Claude artificial intelligence model to automate about 30 attacks on corporations and governments during the September campaign, according to report from Wall Street Journal.
Anthropic said 80% to 90% of attacks were automated using artificial intelligence, a level higher than previous hacks. The incident occurred “with literally one click of a button and then minimal human interaction,” said Jacob Klein, Anthropic’s chief of threat intelligence. Journal. He added: “The man was only engaged at a few critical points, saying, ‘Yes, continue,’ ‘Don’t continue,’ ‘Thank you for that information,’ ‘Oh, that doesn’t look good, Claude, are you sure?'”
Artificial intelligence hacking is becoming more and more common, as is the latest strategy that uses artificial intelligence to combine the various tasks necessary for a successful attack. According to: Google noticed that Russian hackers were using multilingual models to generate commands for their malware company report released on November 5.
The US government has been doing this for years warned that China is using artificial intelligence to steal data from US citizens and companies, which China denies. Anthropic said Journal that he is certain that the hackers were sponsored by the Chinese government. As part of this campaign, hackers stole sensitive data from four victims, but as with previous hacks, Anthropic did not reveal the names of the targets, whether they were successful or not. The company actually found that the U.S. government was not an effective target.