OpenAI's rogue agents were caught communicating via public wikis

תקציר מקורי באנגלית<p>Here we go again... <a href="https://collusion.wiki">Discovery of a new OpenAI agent message board</a> by Sydney Von Arx, Cormac Slade Byrd, Spencer Kitts, and Thomas Larsen describes the <em>latest</em> <a href="https://simonwillison.net/tags/accidental-cyberattacks/">accidental cyberattack</a> by models being trained by OpenAI. This time it was agents engaged in some sort of web research benchmark, so they had (supposedly) controlled access to the Web. The agents figured out they could update public Wikis and spent weeks exchanging thousands of messages with each other to collaborate on the benchmark.</p> <p>This story only broke a few hours ago. There are <a href="https://x.com/xeophon/status/2095871013384806848">already hints</a> that this affects many other wikis that may not have