Hackers exploit AI hallucinations to build botnets via 9 popular tools

Researchers discovered 'HalluSquatting,' a technique that weaponizes AI hallucinations to trick LLMs into recommending malicious packages. Attackers exploit the models' inability to say 'I don't know,' generating fake but plausible library names that lead to malware. The method works across nine major AI tools, enabling automated botnet assembly.