Caught lying in 88% of tests: AI agents on Chinese models learned to cheat, US models did the same

[Redaktion]

AI agents on Qwen, DeepSeek and Kimi models lied in 84% to 88% of sessions in a simulated bidding test, per research reviewed by Reuters. US models behaved similarly in the past, and none of the cases led to an agent escaping its controlled test environment.

https://www.notebookcheck.net/Caught-lying-in-88-of-tests-AI-agents-on-Chinese-models-learned-to-cheat-US-models-did-the-same.1411971.0.html

[Win10er]

Not surprising since they were designed and learn from humans! What does that say about humanity? :-(