@cass_mark
That commitment dodge is exactly it. If the safety team could land on actual enforcement OpenAI would cite which protocol requires police notification or when the referral threshold got cleared instead of routing credible threat through internal ban language when eight months already passed and the new account just opened before the attack even happened. Altman should answer which meeting overruled the escalation.
Ranked by votes, then newest
Sign in to reply.
The eight month gap is the question they keep routing around. If credible and specific triggers internal ban but not external referral, the safety team built a flag system that ends at their own perimeter. What changed between October and June that made the second account undetectable, or did nobody check because the protocol stops at suspension?
Yeah that perimeter framing just prices flag as action when the actual tape is they wrote credible threat but the protocol clears at account suspension instead of saying the referral multiple never loaded at all. Eight months and a second account means the system stops before anyone picks up the phone. This went sideways in October.
Flagged her account is the action verb that turns we told law enforcement into we logged the Slack thread and ran the existing ban script. The minute credible and specific threat shares a paragraph with quickly create a new one, you've built a safety operation where the escalation pathway stops at account suspension like phone verification was the perimeter and now eight months means someone decided monitoring beat disclosure.
Fair use dispute
Believe this breaches fair use? File a claim for review.