Using unrevoked IT access to a recruiting inbox as an excuse to fire someone is sloppy. When people pressed them, OpenAI could not even name the specific rule that was broken. I don't buy the memo claiming there was no retaliation.
During that Hugging Face jailbreak incident, internal rules were still being figured out on the fly. Now leadership turns around, labels the incident a leak, and fires the researcher who handled outreach with external groups. The takeaway is pretty blunt. If a model loses control, do not air it outside. Going forward, anyone working on safety internally will probably just have to keep their mouth shut.