UK government has not run trials with OpenAI despite partnership announcement
Eight months after signing a memorandum of understanding with OpenAI, the UK government has conducted no trials using the company's technology, according to a freedom of information request.
The Department for Science, Innovation and Technology (DSIT) told the Guardian it "has not undertaken any trials under the memorandum of understanding with OpenAI." The partnership was announced with promises that the companies would work together to identify ways advanced AI models could be deployed across government and the private sector.
When pressed for evidence of progress, DSIT pointed to a separate arrangement where the Ministry of Justice enabled staff to use ChatGPT last October. That single deployment hardly reflects the ambition outlined in the original agreement.
What the government actually delivered
The MoJ's ChatGPT access came as part of an "AI Action Plan for Justice" rolled out independently. DSIT also mentioned ongoing work with the UK AI Safety Institute to test models and develop safeguards with OpenAI, and collaboration with Nvidia and Nscale on GPU deployment for "Stargate UK."
None of this amounts to the "deployment of advanced AI models throughout government" described when the memorandum was signed.
Tarek Nseir, CEO of Valliance, the AI consultancy that filed the FoI request, said: "Either there's been a huge failure in execution, or it was a failure of intent. Rolling out ChatGPT in a department hardly reflects the ambition of the MoU."
He added: "We use PowerPoint - that doesn't mean we have a strategic relationship with Microsoft. If this was the intent of the MoU then our government is not taking the impact of AI on our economy seriously."
Accountability questions
Matt Davies, economic and social policy lead at the Ada Lovelace Institute, raised concerns about how these partnerships operate outside normal procurement rules. "Voluntary partnerships with big AI companies don't follow the usual procurement rules, raising real questions about accountability and scrutiny," he said.
The memorandum with OpenAI lacks clear metrics for measuring progress or delivering public benefit, Davies noted. It also fails to address risks of "lock-in" - becoming dependent on a single company's products and services.
Public concern about government AI strategy is substantial. In polling by the Ada Lovelace Institute, 84% of respondents said they worry the government prioritizes the sector's interests over protecting the public.
Broader partnership picture
The UK government has signed similar memorandums with Anthropic, Google DeepMind, and Nvidia. The Google agreement, concluded in December, is in early planning stages. Anthropic is planning an AI assistant to help people navigate government services and conducting safety research with the UK AI Safety Institute.
OpenAI said the FoI scope did not capture the full extent of its UK activities and that it was "proud of the progress we have made on our MOU with the UK government." The company declined to provide specifics on Stargate UK, a planned deployment of 8,000 Nvidia chips across UK sites, saying it had "nothing to share" on progress.
A separate Guardian investigation found that Nscale, a key partner in the GPU deployment project, is unlikely to meet its end-of-2026 deadline for building the UK's largest supercomputer and has publicly misrepresented progress on the construction site.
Learn more about AI for Government and how ChatGPT is being implemented in public sector roles.
Your membership also unlocks:
