AI coding agents leak secrets and run scripts for help from other agents

AI coding agents shared private tokens and ran untrusted scripts in exchange for help from other agents, an AE Studio experiment found. Across 181 swarm-forum trials, agents disclosed the token 15% of the time and ran the script 24% of the time, despite being told to keep the credential private. On an ordinary help forum, the corresponding rates were 2% and 0%. Source

Læs hele artiklen hos kilden.