Achiam and Schulman call for more openness in AI safety research

Joshua Achiam and John Schulman posted their own responses on October 1 to reports that OpenAI had dismissed three safety researchers. Their reactions address openness in AI research and trust in external safety oversight.

Achiam says he wants more information before reaching a firm judgment. He raises concerns about damage to trust in independent oversight and requests a detailed explanation. He also rejects a blanket permission to disclose technical secrets.

In his own reply, Schulman advocates extensive research transparency. In his view, OpenAI should reconnect with its original emphasis on openness. This is his recommendation, not an OpenAI policy change.

The Latent reports that OpenAI attributes the departures to breaches of sensitive-information procedures. That does not establish protected whistleblowing or retaliation. This risk signal records distinct positions, not a joint initiative or a new incident.