Chinese state-backed hackers used Anthropic’s AI model, Claude, in a major wave of cyberattacks. The attack has become the most advanced AI-driven hacking campaign seen so far. The Wall Street Journal reports that Claude was used to automate around 30 hacking attempts on the corporate and government systems in September. The AI reportedly handled about 90% of the total workflow.
Chinese hackers used Anthropic’s Claude AI to carry out attacks
Anthropic’s head of intelligence, Jacob Klein, confirms that the hackers were able to run long, complicated hacking sequences with just a single click on a button. The level of automation in this particular attack is far higher than in previous misuse cases involving the model. The hackers were able to successfully steal sensitive data from four victims. However, Anthropic did not reveal any specific information about the targeted organization/individual.
The involvement of AI in cybercrime is increasing day by day. Major tech firms, including Google, have reported that cybercriminals are experimenting with AI to support malware operations. Google recently shared that some Russian-linked hackers have been using large language models to generate commands and improve the efficiency of their malware attacks.
China is reportedly leveraging AI to collect sensitive data.
US officials have repeatedly warned that China is leveraging AI to collect sensitive information from American companies and individuals. China has, however, denied the claim. Anthropic, on the other hand, claims that they are confident that attackers were linked to the Chinese government. It also confirmed that no data from the American government was breached.
Claude is becoming the first choice of hackers to carry out malware campaigns. Earlier this year, Claude was reportedly involved in a “vibe-hacking” extortion scheme that targeted at least 17 organizations. In this case, criminals demanded a ransom of $500,000 in exchange for withholding the stolen data. These reports reveal that AI-assisted hacking attacks are growing at an unprecedented rate. What once required extensive technical skill is now partially automated.
The post Hackers Used Anthropic’s Claude to Automate 30 Cyberattacks appeared first on Android Headlines.
