This form of artificial intelligence makes access to such operations significantly easier.
Cybercriminals are increasingly using artificial intelligence as a tool. A threat report from Anthropic shows that AI drastically lowers the barriers to carrying out cyberattacks. According to the findings, even perpetrators with limited technical expertise can use AI chatbots such as Anthropic’s Claude to execute complex operations.
In the past month alone, Anthropic reported that 17 companies and organizations from sectors including healthcare, government, and religion became victims of automated attacks. According to the report, Claude was used to identify vulnerabilities and assist in decision-making about how a network could be attacked most effectively and which data should be copied.
Normally, such an operation would have required a team of experts, explained Jacob Klein, head of Anthropic’s Threat Intelligence Team, in an interview with the technology magazine The Verge. Now, it is possible for a single individual to carry out these operations with the help of artificial intelligence. Furthermore, new AI systems can act as “agents” on behalf of users and independently take over tasks to a large extent.
Use in Ransomware Attacks
According to the report, AI chatbots such as Claude are also being deliberately used in data extortion schemes. Anthropic stated that Claude has been used to infiltrate networks, collect and analyze data, and even draft “psychologically targeted” extortion messages. Victims were threatened with the publication of stolen information, and in some cases, ransom demands exceeded 500,000 U.S. dollars.
Klein told The Verge: “This is the most sophisticated use of agents for cyberattacks that I have ever seen.”
North Korean Workers Embedded in U.S. Companies
The report also described other cases where Claude was used in online crime. According to Anthropic, the chatbot was employed by North Koreans who managed to fraudulently secure remote programming jobs in U.S. companies in order to funnel money to their government. They relied on the AI software to communicate with their employers and complete their assignments.
Anthropic’s analysis suggested that these individuals did not possess sufficient software development expertise to complete the work independently without Claude’s support. In the past, North Korea would have had to train such specialists for years, “but with AI, that restriction no longer exists.”
AI Used for Fraud Schemes
Additionally, cybercriminals have used Claude to develop fraudulent models that they offered for sale online. According to Anthropic, this included a romance scam bot on the Telegram platform. The bot was designed to simulate a romantic relationship with victims in order to exploit them financially. Claude was reportedly able to conduct conversations “with high emotional intelligence” in multiple languages.
The company emphasized that although it has extensive measures in place to prevent the misuse of its AI software, cybercriminals continually attempt to bypass these safeguards. Anthropic stated that it intends to use the lessons learned from these cases to further strengthen Claude’s security features.
Source: kim/dpa
Latest Posts
