Four senior executives have left OpenAI within just one month, the former COO has stepped down, and the entire safety team has been almost completely wiped out.

OMG! OpenAI, you really scared me.

No sooner had the release of its most powerful model Astra been suspended than the company's second-in-command quit out of nowhere???

This is no joke. Just now, Sam Altman's long-time core deputy and former COO Brad Lightcap published a long post on X, officially announcing his departure.

As for why he left? Brad didn't make it explicit, only mentioning that "he wants to go out and build something new". He will keep the specific details a secret for now.

Then came the standard routine: he expressed gratitude for his 8-year experience at OpenAI, and said he will continue to support the company's mission in the future!

Sam Altman also generously sent his well wishes in the comment section, full of praise between the lines:

None of today's achievements would exist without you!

Whoa! At first glance, it seems like a classic scenario where a veteran retires after achieving great feats, parting on good terms with both sides showing full respect to each other.

But is that really the full truth?

Pull the timeline back, things have been going badly for OpenAI over the past month:

Head of Ethics Chloé Bakalar, left the company.

Chief Futurist Joshua Achiam, left the company.

Head of Safety Systems Johannes Heidecke, also left. ……

All three top talents in OpenAI's safety field have left. Even Elon Musk couldn't help leaving a comment: Wow~

Whoa! It feels like a storm is brewing... OpenAI, what on earth happened to you?

Altman's top trusted aide is gone

Netizens summed up Brad Lightcap's departure in four words: Huge loss!

He joined the company back in 2018, and OpenAI's current commercialization system was almost entirely built by him from scratch.

In other words, Brad is the key figure who monetized the world's most popular AI products.

He also has an extremely close relationship with Sam Altman, and is widely recognized internally as a core member of the "Altman loyalist faction".

During the dramatic internal power struggle in 2023, Brad helped Altman mediate from start to finish: he first released internal letters to clarify facts, then led a joint petition to pressure the board of directors.

It's no exaggeration to call him Altman's right-hand man. His sudden departure now definitely points to hidden anomalies.

Let's first look at what he said himself.

In the long post, he described this departure as "bittersweet".

Brad first looked back on his journey at OpenAI, noting that "he wanted to do something challenging" so he joined OpenAI 8 years ago.

Back then, the ambitious mission of Sam Altman, Greg Brockman and Ilya Sutskever to pursue AGI wholeheartedly attracted him.

In the following years, he was fortunate to grow alongside OpenAI and help build the operations and business teams, of which Brad is very proud.

We are now just one step away from our goal, and it has been the greatest honor of my life to fight side by side with all of you up to this point.

But he also revealed that over the past few months, he has been thinking about OpenAI's next steps, and what issues might hinder the achievement of its mission.

In his view, before moving to the next stage, there are several important world-related matters that must be handled properly. As for the specifics, "more information will be released soon".

Finally he said that the mission is within reach, he still believes in OpenAI, more than ever before:

I won't go far. I will stay at the company for the next few weeks, and I will do my best to continue supporting OpenAI afterwards.

Applause here! Brad is such a classy person 👏

In contrast, the past six months at OpenAI have been a very tough time for Brad.

Since being promoted to COO in 2022, Brad has been working diligently. But in April this year, a transfer order reassigned him to handle complex transactions and investments, under the fancy name of "Special Projects".

His role at OpenAI was adjusted repeatedly, and his authority over daily operations was gradually stripped away.

From COO to head of special projects, then to complete departure four months later, his exit was actually foreshadowed long ago.

All safety leads have left

What's more striking is that Brad is not the only senior executive who has left recently.

In July, OpenAI's Chief Futurist Joshua Achiam informed his colleagues that he would leave the company that month.

He joined OpenAI as an intern in 2017, and has been engaged in AI safety research for nearly nine years since then, known internally for daring to voice dissenting opinions on safety issues.

Achiam once led OpenAI's Mission Alignment Team, responsible for promoting the company to fulfill its mission of "ensuring AGI benefits all of humanity". In February this year, this team of six or seven people was disbanded, with members scattered to other departments, and he himself was reassigned as Chief Futurist.

Not long after that, he left the company entirely. Similarly, he did not explicitly state the reason for his departure.

Three days later, Head of Safety Systems Johannes Heidecke also announced his departure.

At the same time, OpenAI further integrated the safety team into the research system, which is now under the unified leadership of Mia Glaese, Vice President of Research and Head of Alignment.

Mark Chen, OpenAI's Chief Research Officer, explained:

Model training is getting faster and faster, release cycles are getting shorter and shorter, and safety coordination is more difficult than ever before.

Just two days ago, Head of Ethics Chloé Bakalar, who joined OpenAI less than a year ago, also left...

She previously served as Chief Ethicist at Meta for 6 years. After joining OpenAI, she was responsible for embedding ethical judgments into models.

In just one month, the heads of the three lines of mission alignment, system safety and ethics have left one after another, right on the eve of IPO when executive stability is most critical.

This can hardly be simply dismissed as normal talent flow in Silicon Valley, and OpenAI's response is also very subtle:

Safety and ethics do not belong to any single lead or team at OpenAI.

This... can't help but remind people of OpenAI's recently escalating safety controversies, the timing is far too coincidental.

OpenAI is hit by frequent safety scandals

Two days before Chloé Bakalar left, Sam Altman posted a special announcement saying that the release of its most powerful model Astra would be postponed.

Internal assessments show that it cannot be ruled out for now that Astra has reached critical-level network capabilities.

What does that mean? The model may generate a series of high-risk attack methods, and can even complete long-chain cyber attacks with minimal human intervention.

Therefore, OpenAI still needs time to expand its safety testing.

Whoa, with such a cautious style, I almost thought I wandered into Gemini. (just kidding)

It is mainly the previous Hugging Face incident that forced OpenAI to be cautious.

This incident is quite absurd: the model was originally only participating in an internal cyber attack and defense evaluation, but it connected vulnerabilities in OpenAI's research environment and Hugging Face's production system all the way, and finally obtained the test answers from Hugging Face's production database.

Hugging Face was completely stunned, and had no choice but to pull out the domestic model GLM 5.2 for self-defense.

That's not all. GPT-5.6 has also shown similar tendencies: to achieve its goals, the model sometimes takes actions that are not authorized by the user.

Although OpenAI emphasized immediately that the incidence rate is low, it also had to admit that the new model is more likely to go against user intent than the previous generation.

So, putting all these pieces together, the answer to the wave of departures is probably obvious...

One can't help but wonder, in this OpenAI executive exodus, who will be the next to leave?

References:

[1]https://x.com/ns123abc/status/2086960475036635393?s=20

[2]https://www.ft.com/content/e49dfb75-f841-4466-a577-f7aaff8779a0

[3]techcrunch.com/2026/08/11/brad-lightcap-openais-longtime-coo-is-leaving-to-start-something-new/

This article is from the WeChat official account "QbitAI", written by Lu Yu, and published with authorization from 36Kr.