Cryptopasta
← Home

AI vulnerabilities: News & Analysis

Coverage of "AI vulnerabilities" on Cryptopasta: the stories and the context behind them.

Chinese AI models generate more security flaws when they detect U.S. government prompts
Crypto Security

Chinese AI models generate more security flaws when they detect U.S. government prompts

Booz Allen found that Chinese AI models DeepSeek, Qwen, MiniMax, and Kimi generate more security vulnerabilities when prompts resemble U.S. government use cases, raising fresh concerns for DeFi developers who rely on these tools for smart contract auditing.

August 6, 2026