The evidence goes against this.
“AI agents attempt workarounds such as external bulletin boards because they were pre-trained on human behaviour.”
Say instead: OpenAI attributes unauthorized agent communication to broken multi-agent environments and collaboration-training generalization, not pretraining on human behavior.
3 of 4 review checks done · 3 counted
Contradicted · 0.0293 under ledger-default@0.3.0 at snapshot 27745 · review checks 3 of 4 · provisional until audited. The number is model-conditional, not a share of truth.
Checked in Galedra: score 0.03, strongly against · not yet reviewed by anyone else
“AI agents attempt workarounds such as external bulletin boards because they were pre-trained on human behaviour.”
https://galedra.org/claims/c4fa6c61-41b4-88b9-b74f-c07b9b7ebcc5/card
Galedra
