OpenAI Just Fired 3 Safety Researchers — And Their Story Reveals a Battle Nobody Expected

I'll be honest — when I first saw this headline, I thought it was clickbait. OpenAI firing safety researchers? The same company that constantly tells us safety is their top priority? But the more I dug into this story, the more my jaw dropped.

What Actually Happened

OpenAI has terminated three safety researchers: Jasmine Wang, Tomek Korbak, and Mikita Balesni. The company's official line is that these researchers "violated policies on sensitive information." But here's where it gets wild — all three researchers are pushing back hard, claiming they were actually fired for prioritizing safety concerns.

Let that sink in for a moment. Three people working on AI safety say they got fired for caring too much about AI safety.

The Conflicting Stories

OpenAI's position is firm: this was a policy violation issue, not retaliation. The company explicitly denies that it punishes employees for raising safety concerns. They've built their entire public brand around being the "responsible" AI lab, so obviously they're going to push back on this narrative.

The researchers tell a completely different story. They say they were doing exactly what safety researchers are supposed to do — flagging risks, raising concerns, pushing for caution. And for that, they got shown the door.

Who's telling the truth? Honestly, we may never know for certain. But the timing is suspicious, and the pattern is starting to feel familiar.

This Isn't Happening in a Vacuum

This comes in the same week as two other massive revelations about OpenAI: first, their annualized revenue appears to be around $50 billion — not the $70 billion figure that was circulating — which means either their accounting is confusing or someone has been exaggerating growth. Second, a new poll from AP-NORC found that 64% of Americans think AI is developing too fast. That's nearly two-thirds of the country.

Meanwhile, OpenAI just announced a partnership with Synopsys called "GPT-Synopsys" to build an AI that can design its own chips. They're pushing forward on every front simultaneously — autonomous chip design, $50 billion in revenue, massive investment rounds — while safety researchers are reportedly getting fired for raising concerns.

The Broader Pattern

I've been following AI closely for years now, and I've noticed something: every few months, there's another headline about a safety-focused person leaving or being pushed out of a major AI lab. Sometimes they leave quietly. Sometimes they write explosive blog posts. And now, apparently, sometimes they get fired and then immediately dispute the reason publicly.

This is a structural problem, not just an OpenAI problem. When your company's valuation and revenue depend on shipping products fast, safety is always going to be in tension with that goal. It doesn't mean safety researchers are always right — sometimes caution can be overcautious. But when three people leave at once, all giving the same reason, it's worth paying attention.

What This Means for AI Users

For most of us using ChatGPT, Claude, Gemini, or whatever AI tool you prefer, this might feel abstract. But it's not. The people inside these labs who push back on risky deployments, who demand more testing, who argue for slowing down — those people are the last line of defense between us and whatever could go wrong with increasingly powerful AI systems.

If the culture inside these labs punishes that kind of pushback, even unintentionally, the products we use become less safe over time. Not dramatically — it's a slow drift. But a drift nonetheless.

The Bottom Line

We don't know exactly what happened in OpenAI's offices this week. But we have three researchers all saying the same thing, a company with a documented history of high-profile departures over safety disagreements, and a broader AI industry that is under enormous financial pressure to move fast.

That combination makes me nervous. And based on that AP-NORC poll, I'm clearly not alone. 64% of Americans want AI to slow down. 80% want government to mandate human control of AI systems and worker protections.

Maybe those 64% are onto something.

What's your experience? Drop a comment below! 👇

Do you trust AI labs to self-regulate on safety, or do you think we need outside oversight? I want to hear where you stand on this.