Will OpenAI’s New Safety Team Really Secure ChatGPT?

OpenAI is forming a new safety team led by CEO Sam Altman, with board members including Adam D'Angelo and Nicole Seligman. The committee will advise on key safety decisions for OpenAI projects and operations, addressing concerns raised by several AI researchers who left the company this month.

The team's primary task is to evaluate and further develop OpenAI's processes and safeguards, then submit its findings to the board.

All three leaders of the safety team sit on the board, after which the board will decide how to implement the team's recommendations.

The team was created after co‑founder and chief scientist Ilya Sutskever departed; he had supported the board's attempt to remove Altman last year and co‑led OpenAI's Superalignment team, which aimed to guide and control AI systems smarter than humans.

Another co‑leader of the Superalignment team, Jan Leike, also resigned shortly after Sutskever's departure, stating that OpenAI's safety had given way to flashy products. OpenAI subsequently disbanded the Superalignment team. Last week, policy researcher Gretchen Krueger also resigned for similar safety concerns.

In addition to the new safety committee, OpenAI announced it is testing a new AI model, without confirming whether it is GPT‑5.

Earlier this month, OpenAI released a new ChatGPT voice called Sky that sounded remarkably like Scarlett Johansson. Johansson later confirmed she declined Altman's proposal to voice ChatGPT. Altman subsequently said OpenAI never intended Sky to sound like Johansson and that he contacted her after a voice actor had been selected. The incident raised concerns among AI fans and critics.

Other members of the new safety team include lead Aleksander Madry, safety lead Lilian Weng, calibration science lead John Schulman, safety lead Matt Knight, and chief scientist Jakub Pachocki. However, because two board members (and Altman himself) lead the new safety committee, OpenAI does not appear to have fully addressed former employees' worries.

OpenAI disclosed it has begun training the next frontier model, which is expected to bring higher capability on the path toward artificial general intelligence (AGI).

Subsequent agreed recommendations will be shared publicly in a safety‑compliant manner.

The committee's members include chairman Bret Taylor, Quora CEO Adam D'Angelo, former Sony Entertainment executive Nicole Seligman, and six OpenAI employees such as Sam Altman and co‑founder John Schulman. OpenAI says it will also consult external experts during the process.

"I will reserve my judgment, wait for OpenAI's adopted recommendations to be published, and see how they are implemented, but intuitively I do not fully trust OpenAI (that while trying to win the AI race, they will prioritize safety and ethics)."

This is a disgrace. In an ideal world, all tech companies, regardless of AI competition, should place ethics and safety on equal footing with innovation.

We hope this person's proof is wrong.

How do you see it? Feel free to comment.