AI model turns into cybercriminal tool: What it means for security

Anthropic's report reveals criminals using Claude AI for ransomware, fraud, and data theft - enabling one operator to act like a full cybercrime team with AI-powered attacks

Anthropic, the company behind the large language model Claude, has released a report detailing how people have misused the artificial intelligence (AI) platform with “criminal intent”. It cited various cases of misuse, including a large-scale extortion operation, a fraudulent employment scheme from North Korea, and the sale of AI-generated ransomware by a cybercriminal with only basic coding skills. 

What’s happening

 

*  AI models are being used to perform sophisticated cyberattacks, not just advise on how to carry them out

  *  Actors with limited technical skills have used AI to conduct complex operations, like developing ransomware

  *  Cybercriminals have embedded AI for various operations, including profiling victims, analysing stolen data, stealing credit card information

 

What’s the implication

 

-  A single operator can achieve the impact of an entire cybercriminal team with AI assistance

  -  AI makes both strategic and tactical decisions about targeting, exploitation, and monetisation

  -  Defence becomes difficult as AI-generated attacks adapt to defensive measures in real time

 

Criminal proceeds

 

*  17 organisations targeted by one cybercriminal in August by using Claude Code

  *  $500,000+ paid byorganisations to that criminal who used Claude Code in an ‘unprecedented’ manner

  *  Crime resulted in the compromise of personal records including health care data, financial information, government credentials

 

Source: Anthropic’s ‘Threat Intelligence Report: August 2025’