Disclosure: I run OpenJobs AI. We do AI recruiting. OpenAI just announced they're building the same thing. I have a lot of feelings about this and not all of them are rational.

September 4, 2025. I was in a meeting—one of those pointless syncs where everyone's on mute doing email—when my phone started buzzing. Slack. Text. Another text. Then my co-founder calling.

"Did you see it?"

"See what?"

"OpenAI. Jobs platform."

I pulled up the announcement. Read it. Read it again. Went to get coffee. Came back. Read it a third time.

My first thought: well, shit. Second thought: actually maybe this validates what we're doing? Third thought: no we're fucked. Fourth thought: wait, are we though?

I've been going back and forth on this for three months now. I still don't have a clean answer. But I've got some thoughts.

What They Announced

Fidji Simo—she runs applications at OpenAI—put out a blog post. "Expanding Economic Opportunity with AI." Very White House friendly language. The substance: they're building a jobs platform, launching mid-2026, plus a certification program to train 10 million Americans in AI skills by 2030.

Partners include Walmart, John Deere, Accenture, BCG. And Indeed, which is interesting—I'll get to that.

Josh Bersin, who's been writing about HR tech since forever, pointed out that "this is the same language LinkedIn used in 2008 when they launched Recruiter." He wasn't being nice about it.

The Obvious Irony

Look, I have to say it even though everyone's thinking it: the company whose technology is eliminating white-collar jobs is now building a platform to help people find work.

The arsonist selling fire extinguishers.

I'm not judging. I'm building AI recruiting tools too. We're all in the same weird moral position. But come on. There's a white-collar recession happening. ADP says jobs requiring advanced degrees are now the slowest-growing category. College grads are the least "in-demand" cohort. Meanwhile retail and food service wages are growing 3x faster than knowledge worker pay.

GPT did a lot of that. Or accelerated it. And now OpenAI wants to help you deal with the aftermath.

Is this corporate responsibility? A massive business opportunity? Guilt? All three? I genuinely don't know.

The Microsoft Thing

Okay so Microsoft put $14 billion into OpenAI. Microsoft also owns LinkedIn, which they paid $26 billion for back in 2016. LinkedIn does like $16 billion a year in revenue and Talent Solutions is a big chunk of that.

OpenAI is building a direct competitor to one of Microsoft's most profitable businesses.

The relationship has been getting weird. Microsoft called OpenAI a competitor in their annual filing last year. There were reports that OpenAI wouldn't share technical details about o1 with them. Microsoft's apparently been testing models from xAI and Anthropic and even DeepSeek as backups.

Reid Hoffman—LinkedIn's co-founder—was an early OpenAI investor. This whole thing is a mess of overlapping interests.

But honestly? I don't think Microsoft can do anything about this. OpenAI is too big now. The investment bought Microsoft technology access and some board influence, not product control. Sam Altman does what Sam Altman wants. We saw that in November 2023.

What They're Actually Building (As Far As I Can Tell)

Details are thin because this thing doesn't launch for another year and a half. But:

Matching that actually works. Not keyword matching. Not the "AI-powered" matching that's really just keyword matching with a neural network wrapper. Actual semantic understanding. You describe what you need, the system gets what you mean, matches on real capability instead of resume buzzwords.

Every ATS vendor claims to do this. None of them really do. The difference is OpenAI actually has the technology. Whether they can translate that into recruiting outcomes is a different question.

Certification tied to hiring. Four tiers—basic AI literacy up to expert prompt engineering. You learn through ChatGPT, get certified, those credentials factor into matching. It's this closed loop where OpenAI teaches you AI, certifies that you know AI, then matches you with jobs that need AI. I can't decide if this is genius or dystopian or both.

The Indeed partnership. This is the part everyone's sleeping on. Indeed is one of the biggest job sites in the world. Having them as a launch partner means OpenAI gets job listing data and candidate flow from day one. It also means Indeed is hedging—they're not sure if OpenAI is going to be friend or enemy.

Why It Might Actually Work

I spent five years at BOSS Zhipin and Liepin. I've seen what wins in recruiting tech.

(Side note: there was a cleaning lady at BOSS Zhipin who remembered every single employee's name. Thousands of us. She'd greet you by name in the hallway at 7am. I still don't know how she did it. We hit 200 million users that year. I don't know what that proves but I think about it sometimes.)

Anyway. Here's why OpenAI might pull this off:

The technology gap is real. Most "AI recruiting" tools are garbage. BERT wrappers on resume parsing. Slightly smarter filters. OpenAI is 2-3 generations ahead of what recruiting vendors are using. That matters.

The certification play is clever. If 10 million people get OpenAI certified and employers start preferring those credentials, OpenAI has built a massive candidate base without spending anything on user acquisition. Walmart signed on. Largest private employer in America. Once Walmart is hiring based on OpenAI certs, everyone follows.

LinkedIn is genuinely hated. Recruiters complain about spam and noise. Candidates complain about ghosting. The feed is garbage. The core value prop—connecting professionals with opportunities—got buried under engagement optimization years ago. If OpenAI builds something clean and focused, people will switch. Happily.

Why It Might Not

But also:

LinkedIn has a billion members. Twenty years of career data. Skills endorsements, connection graphs, messaging patterns, who's looking, who's not, career trajectories. OpenAI has ChatGPT conversation logs. Not the same thing.

Network effects are brutal. Two-sided marketplace dynamics. Employers only show up if candidates are there. Candidates only show up if jobs are there. LinkedIn solved this over two decades. Google tried Google for Jobs. Flopped. Facebook tried professional networking. Flopped. This problem is harder than it looks from the outside.

Every recruiting platform in history has claimed better matching. I'm serious. Go look at the marketing copy for any ATS built in the last ten years. "AI-powered matching." "ML-enhanced candidate discovery." "Semantic understanding." They all say it. Most of them are lying. OpenAI actually has the tech—but translating language model capability into hiring outcomes isn't straightforward. Nobody's figured it out yet.

And mid-2026 is a long time. LinkedIn is already integrating AI across their products. They just updated their privacy policy to train models on user data by default. By the time OpenAI launches, LinkedIn will have closed a lot of the gap. First-mover advantage in AI doesn't last when everyone can access the same foundation models.

The BOSS Zhipin Lesson

Let me tell you about something I watched happen.

When I joined BOSS Zhipin we were tiny. Zhaopin and 51job dominated. Our thesis was simple: let candidates and hiring managers chat in real time. Skip the application form. Skip the recruiter. Just talk.

We grew from 50 million to 200 million users in a few years. And here's the thing—our technology wasn't dramatically better than theirs. Not initially. What we did was change the interaction model. We made recruiting feel different.

I think about this when I look at OpenAI's play. Their matching algorithms are probably 2x better than LinkedIn's. That's not enough. 2x doesn't overcome network effects. But if they can make job search feel fundamentally different—conversational, responsive, like talking to someone who actually gets what you're looking for—that might be enough.

Imagine describing what you want to ChatGPT and having it actually understand. Not keyword boxes. Not "upload your resume and pray." A real dialogue. That's different enough that it might work.

Or it might not. I really don't know.

What This Means For My Company

Okay, here's where I have to be honest about my bias.

OpenJobs AI is building in this exact space. Our angle is what I call the Reachability Graph—we optimize not just who to contact but how and when and through what channel. Email deliverability. InMail timing. Phone pickup rates. Our bet is that matching isn't the bottleneck. Getting candidates to respond is.

Does OpenAI's entry validate that thesis? Sort of? They're saying "AI-native recruiting is the future," which helps everyone in the category.

Does it threaten us? Yeah. Obviously. If they build matching AND engagement AND scheduling AND assessment, smaller players get crushed.

My hope is that they focus on matching—that's their strength—and leave the operational complexity to specialists. Integrating with fifty different ATS platforms, managing multi-channel outreach, handling compliance in 200 jurisdictions... that's not what OpenAI is good at.

But maybe I'm just telling myself a comforting story. Cope, as the kids say.

Random Other Thoughts

Candidates don't trust AI hiring. 71% of Americans oppose AI making final hiring decisions. 66% say they'd avoid applying to jobs that use AI in the process. Only 26% trust AI to evaluate them fairly. OpenAI has a marketing problem as much as a technology problem.

The EU AI Act is now in effect. HR AI is "high-risk." NYC requires bias audits. Regulation is tightening, not loosening. Mid-2026 might be launching into a headwind.

Here's something nobody's talking about: AI skills command a 56% wage premium right now partly because they're scarce. Only 12% of workers did any AI training last year. OpenAI wants to certify 10 million people. What happens to the premium when AI skills aren't scarce anymore? Supply up, premium down. The "economic opportunity" they're selling might be more limited than current data suggests.

(I should probably organize these thoughts better but I've rewritten this section three times and it keeps getting worse so I'm just leaving it.)

Predictions

I'll put some stakes in the ground. Check back in 2027.

They launch late. Mid-2026 is aspirational. Building a recruiting platform at scale is operationally brutal. LinkedIn took six years from founding to launching Recruiter. I'm betting late 2026 at earliest, probably 2027, and it's more limited than the announcement implied.

Certification is the real product. The education angle is the stronger play. Teaching AI, certifying people, building a credentialed workforce—that makes sense. The matching platform is harder. I think they nail certifications and struggle with matching.

Consolidation happens. OpenAI's entry validates the category. Bigger players start acquiring smaller ones. The fragmented market of point solutions collapses. Some of us get bought. Some of us die. That's just how it goes.

Trust stays low. Candidates remain skeptical. The distrust doesn't get solved by 2026. Maybe not ever. Human oversight in hiring stays the norm because candidates demand it.

What Should You Actually Do

If you're looking for work: learn AI skills. The wage premium is real today. Get certified somewhere—OpenAI, Coursera, Google, whoever. Build a portfolio that shows what you can actually do. If matching moves from keywords to capability, tangible evidence matters more.

If you're an employer: don't panic. This isn't launching for 18 months. LinkedIn isn't going anywhere. But start thinking about skills-based hiring. Write job descriptions that describe what you actually need, not credential proxies.

If you're building in recruiting tech: good luck to all of us. Find a niche. Specialize. Or prepare to get acquired.

The Thing I Keep Coming Back To

OpenAI's technology is eliminating jobs. Their platform promises to help people find new ones. But what if there aren't enough new ones?

The white-collar recession might be structural. AI can do knowledge work cheaper and more consistently. That demand might not come back. Training 10 million people in AI skills doesn't create 10 million AI jobs.

Maybe that's not OpenAI's problem to solve. Maybe it's unfair to expect them to fix labor market disruption that their tech is causing. Maybe the best anyone can do is provide tools for adaptation.

But let's not pretend this is charity.

Recruiting is a hundred-billion-dollar market. LinkedIn Talent Solutions alone does billions. There's money here. A lot of it. The "economic opportunity for all" framing is nice. The White House partnership is great PR. This is also a business move.

The company whose technology threatens your job is now offering to help you find a new one.

I don't know what to make of that. I'm not sure anyone does.