AI ethics roles at frontier labs are hollow
OpenAI's head of ethics left less than a year after joining. She came from Meta, where she held the same title. The thread was blunt: commenters argued there is no one actually responsible for ethics at OpenAI in any meaningful way, and that the role itself is more about optics than governance.
This lands alongside the reasoning trace vulnerability story and the Grok Bot launch (xAI opening its model as an API product), and the picture that emerges is of an industry that has largely stopped pretending the safety and ethics infrastructure matters. OpenAI specifically has cycled through high-profile safety and ethics departures repeatedly, and each time the company gets bigger and the departures get quieter.
The counterpoint some made: maybe these roles are structurally impossible to do well inside a company racing to ship. The incentives do not reward slowing down, so the people hired to slow things down leave.
So what?
For founders pitching enterprise customers or government contracts, the optics of AI ethics and governance matter even if the substance does not. The steady exodus of credible people from these roles at big labs creates both a risk and an opportunity: a risk if your product is associated with the same governance theater, and an opportunity if you can demonstrate actual accountability mechanisms.