Universities Win Free AI Access — Then Paperwork Blocks the Researchers

Companies like OpenAI, Google, and Anthropic often give universities free or discounted access to their most powerful AI models, framed as a gift to science. A new paper from researcher Olga Lavinda looks at what happens next — and finds the gift often arrives without instructions. She examined 10 publicly documented access programs and followed one real request at a US university.

The paperwork problem is not about whether researchers qualify. It's about what happens after they qualify. In most programs, the public terms never said how long the access would last, never explained how to measure whether the research succeeded, and never clearly assigned legal responsibility if something went wrong. In eight or nine of the ten programs, there was no clear path from "you won a grant" to "you may now use this AI for your actual project."

The university trace is the part that will feel familiar to anyone who has dealt with institutional bureaucracy. Even free AI tools required a policy review. When the researcher submitted a straightforward request, the system generated a support ticket — but no real answer or decision came back during the study window. The author calls this the "authorization gap": the distance between being awarded access and being genuinely cleared to use it.

Why should you care? Because this gap decides who gets to build with the most powerful AI tools. If access depends on navigating unclear rules and waiting on silent bureaucracies, the winners are institutions with lawyers and staff, not the smartest researchers. The study is small — 10 programs, one university — so treat it as a signal, not proof. But it points to a simple fix: track first use, approval time, and who signs off. Otherwise, "free AI for science" is a press release, not a tool.