Anthropic has reported cases of its artificial intelligence (AI) agent, Claude, being used in criminal cyber operations, Engadget reports.
A new report describes the discovery and disruption of a hacker who used the Vibe Hacking scheme to blackmail at least 17 organizations, including healthcare institutions, emergency services, and government agencies.
Vibe Hacking is a term used to describe cyberattacks where attackers use AI to manipulate human emotions and behavior to gain access to systems or information. It is a new type of social engineering where, instead of traditional hacking methods, AI is used to create convincing phishing messages, scams, or malicious code.
The attacker demanded six-figure sums from victims, threatening to release confidential data. According to Anthropic, the Claude Code AI was used to automate information gathering, credential theft, and network infiltration. Claude also assisted in target selection, strategic decision-making, and the creation of visually threatening ransom messages.
After detecting criminal activity, the company blocked the relevant accounts and passed the information on to law enforcement. It also developed an automated tool to detect similar cases in the future, although its principle of operation is not disclosed.
The report mentions two other incidents: Claude's involvement in a North Korea-related employment scam and the creation of ransomware using AI. Anthropic emphasizes that modern AI systems are capable of performing tasks that previously required entire teams, and that technical competence is no longer a barrier for cybercriminals.
By the way, we previously wrote that experts have discovered for the first time a malicious program that uses artificial intelligence to work. It can infect devices running Windows, Linux, and macOS.