On the site for its new certification programs, OpenAI has laid out a plan larger than certificates: to combine training, credentialing and access to economic opportunity into one legible route, which the company says will form the basis of a future OpenAI jobs platform. Ella Hafermalz, associate professor of work and technology at the KIN Center for Digital Innovation at Vrije Universiteit Amsterdam, argues that universities are watching the wrong thing. They are busy adapting to students using AI, she writes, while the more consequential move is an attempt by the largest AI companies to occupy the entire path from coursework to a first job.
Her case rests on three pieces that OpenAI has assembled separately and that only look like a strategy when you put them side by side.
The first is access to students. OpenAI is recruiting campus representatives through a program called Student Collective, whose listing promises prospective campus leaders training on OpenAI tools, funding, branded merchandise and access to a community of exceptional peers. Hafermalz calls this a familiar brand strategy: presence inside a community buys trust and influence, and generates data that can shape what the company does next.
The second is access to employers. OpenAI Deployment Company launched this year with more than $4 billion in initial investment. It places more than 150 "forward deployment engineers" — a military metaphor for specialists who work inside client organizations — across its network of companies, where their job is to rebuild workflows around OpenAI's tools. A company helping to reshape work across several industries at once both sees and partly determines what kinds of workers employers need.
The third is the credential. Working with partners including Accenture and John Deere, OpenAI offers certificates meant to help graduates find jobs; the current programs center on AI tools and digital literacy. Training people to use your own products is not new, as Hafermalz concedes — technology companies have always done it. What is different is where the certificates sit. OpenAI describes the program as full instruction inside ChatGPT: learners practice on real tasks, get contextual feedback and review their own work in a single environment, with ChatGPT acting as teacher, practice space and feedback mechanism at once.
Put together, that is a single service in which a young person learns, is certified and is matched to work, with the same company's tools underneath every step.
Hafermalz's objection is not that this would fail. It is that it might work. Students are anxious about the future and employers are urgently looking for people with the right skills, so the offer has real appeal. But handing the whole path to one centralized system carries human, social and potentially economic costs. She points to a 23-year-old who recently told the Wall Street Journal that the appeal of "internships for office professionals" was that companies want to help shape employees around their own needs. Useful for the employer. Her question is what happens to the next generation's ability to work out who it wants to be, and what exists beyond the current requirements of these structures — and what happens when employer demand shifts because the technology moves on, or after the next market bubble bursts.
What universities should do instead, in her account, is stop treating AI as a way to grade faster or as a substitute for teaching staff and concentrate on what OpenAI cannot easily supply: independence, access to expert knowledge in the right context, productive difficulty, and social ties. The formative moments, she writes, come when peers and teachers see a student's effort and notice both who they are and who they are becoming.
The sequencing is the strongest part of her argument, and she states it plainly. Sam Altman and his peers have spent recent months warning that jobs are about to disappear. The same firms are now selling one clear path through the disruption they described. That is worth naming even if every individual claim in it is sincere.
The detail I would put more weight on than she does is the Deployment Company. A vendor whose engineers are rewriting workflows inside employers, and who also trains candidates, certifies them and intends to run the hiring platform, is setting the requirements and grading the answers. In most industries that arrangement has a name and a disclosure requirement. In credentialing it is called vertical integration and gets a product page. Notably absent from OpenAI's announcements is any evidence about what the certificates are worth in the labor market — no placement rates, no employer commitments beyond the two named partners, nothing an applicant could use to compare the credential against a degree.
The question neither side is asking is who accredits the accreditor. Universities are slow, expensive and frequently bad at connecting graduates to work, which is precisely the opening here. But their credential is checked by somebody other than themselves, and it is portable across employers who do not use a particular vendor's tools. A certificate earned inside ChatGPT, validating skills defined partly by OpenAI's own deployment engineers, is worth exactly what OpenAI's hiring platform says it is worth.
The wager universities are being asked to make, mostly without noticing they are making it, is that employers will keep valuing a degree for reasons a software company cannot replicate. The 150 forward deployment engineers are already inside those employers, rewriting the jobs the degree is supposed to qualify people for.