OneGov AI Deal Ends, Testing Federal Workers' Habit Lock-In

What happened

The US federal government's OneGov bulk-licensing arrangement, which gave agencies access to ChatGPT, Gemini and Claude, is coming to an end — leaving federal employees who have built these tools into their daily workflows facing decisions about whether to renew, switch providers or lose access altogether, according to FedScoop.

The OneGov framework had allowed agencies to procure generative AI tools at scale rather than negotiating individually, embedding the three major chatbot platforms into routine government work. With the deal expiring, that centralised access model is now in question, and workers who have come to rely on these assistants must navigate what happens next without the same coordinated procurement backing them.

Why it matters

This is less a story about a single contract lapsing and more about what happens when large-scale AI adoption inside an institution reaches the point where habitual use outpaces the procurement cycle that enabled it. Once employees have restructured how they draft, research, summarise or code around a specific assistant, the tool becomes infrastructure rather than an optional add-on — and its withdrawal or replacement carries real operational friction, not just a licensing footnote.

For public-sector digital transformation more broadly, the episode is a useful signal: bulk AI licensing programmes accelerate adoption faster than most organisations plan for the exit, renewal or migration scenarios that inevitably follow. Agencies now face a practical governance question — how to manage tool continuity, retraining and data portability when the underlying commercial terms shift — that most transformation roadmaps have not yet addressed in detail.

The Renascence take

The interesting part of this story isn't the contract expiry — it's the behavioral trap it exposes. When an organisation rolls out a tool at scale without a clear plan for what happens after the introductory or bulk-procurement period, it is effectively running an uncontrolled experiment in habit formation, and the "control group" is whoever has to manage the fallout.

Bulk AI deals are designed to drive adoption, and adoption is designed to become habit — but habits don't renegotiate contracts, procurement teams do. The real lesson for any organisation rolling out AI at scale, public or private, is that the switching cost isn't technical, it's behavioral: once people have rebuilt their workflow around a specific assistant's quirks and outputs, "just use a different model" is not a neutral instruction. Leaders who want durable AI adoption need to design for the renewal conversation from day one — not treat it as an IT afterthought once the invoice is due.