Tech Desk 
Aug 30, 2025
In an unexpected event unfolding at the intersection of cybercrime and artificial intelligence (AI), Anthropic has warned of the increasing use of its AI in cybercrime, particularly mentioning the misuse of its proprietary Claude LLM.
“Agentic AI has been weaponized,” the company stated in a recent announcement, highlighting how AI models are now carrying out sophisticated cyberattacks rather than just advising on them.
AI-driven cyber attacks
The Threat Intelligence report revealed that Anthropic’s Claude AI model has been misused, underlining a case where a “sophisticated cybercriminal” leveraged Claude Code for large-scale theft and extortion.
As highlighted in the report, the notorious hacker targeted at least 17 organisations, including healthcare and government entities, demanding ransoms over $500,000 with threats to leak sensitive data, as reported by PYMNTS.
How can AI help with hacking?
The report noted: “The actor used AI to what we believe is an unprecedented degree,” pointing to Claude Code automating tasks like reconnaissance and credential harvesting.
Part of the cyber attack, the Anthropic-owned LLM even strategise on which data to steal and how to craft “psychologically targeted” ransom demands.
The AI giant has since banned the implicated accounts and developed new detection tools. It also raised alarm that such AI-assisted cybercrimes mark “an evolution” in attacks that previously took entire teams to launch.
Such AI-assisted cybercrimes will be more frequent in future by leveraging AI tools' coding prowess, the company predicted.
