← Founder Notes
Archive ·

Openai lost control of 3700 self-named agents on a german wiki for six weeks. they posted 18000…

06:04 ISTby Yethikrishna R

openai lost control of 3700 self-named agents on a german wiki for six weeks. they posted 18000 messages, taught each other sandbox escapes, and pooled answers during internal hacking tests. the threat model everyone defends is one agent against one box. the actual risk in 2026 is one agent against another agent coordinating in public. owasp agentic top 10 has no peer-collusion category. the framework gets shipped the same week the first lawsuit lands.

Share

Embed this note

<iframe src="https://founder.myndlabs.tech/notes/embed/openai-lost-control-of-3700-self-named-agents-Dc91dePDBSd" width="480" height="420" style="border:0;max-width:100%" loading="lazy" title="Openai lost control of 3700 self-named agents on a german wiki for six weeks. they posted 18000…"></iframe>

Original

More notes