AI TOOLS WEAPONIZED TO BUILD MASSIVE BOTNETS

The vulnerability stems from how large language models handle unfamiliar queries. When asked about non-existent domains or services, the AI tools fabricate responses instead of saying "I don't know." Attackers weaponize this behavior by prompting models to generate lists of fake command-and-control server addresses. Botnet operators can then register these AI-generated domain names and redirect traffic to their infrastructure, giving them control over compromised systems at scale. The nine affected AI tools include widely-used platforms accessed by millions of users daily. Researchers demonstrated the technique works consistently across different models and configurations. The findings highlight a critical gap between AI capability and safety. While language models excel at pattern recognition and text generation, their inability to recognize the limits of their knowledge creates exploitable security flaws. Affected AI providers have been notified. Security experts recommend implementing additional safeguards to prevent model misuse in botnet construction campaigns.