Anthropic Says More Than 60% of Banned Accounts Used AI to Prepare Cyberattacks

Generative AI has lowered the barriers to malware development and vulnerability reconnaissance, allowing attackers with limited technical skills to plan intrusions. Anthropic has continued to track misuse of Claude. Because similar technology can also identify smart-contract weaknesses, it is increasing security risks for DeFi and crypto assets.

Anthropic’s latest report said about 67% of accounts banned in the year before its publication had used AI to write malware or prepare cyberattacks. AI-assisted attackers rated at “moderate” risk or higher also accounted for more than 50%. Experts warned that AI’s ability to analyze smart-contract vulnerabilities is placing direct pressure on DeFi defenses.