Inspired by the Guardian article on which Tarek Nseir was asked to comment: https://www.theguardian.com/politics/2026/mar/21/uk-government-yet-to-trial-openai-tech-months-after-signing-partnership
Nothing in the agreement required a trial. The memorandum of understanding the UK signed with OpenAI in July 2025 names three areas of cooperation and states that it is voluntary and not legally binding. It sets no timeline and names no first use case. When Valliance asked the Department for Science, Innovation and Technology what had been trialled under it, the department confirmed that nothing had.
What the FOI request found
Valliance submitted the request eight months after the memorandum was signed. DSIT confirmed it “has not undertaken any trials under the memorandum of understanding with OpenAI”. The Guardian reported the finding on 21 March 2026.
The department said work under the agreement was continuing, and pointed to a separate initiative from October 2025 that gives civil servants access to ChatGPT with UK data storage. The memorandum did not create that rollout, and it falls outside the areas the memorandum covers. OpenAI said the response did not reflect the full scope of its work with the UK.
“There’s no shortage of ambition or spending in the UK, but in truth the issue is in delivery.” — Tarek Nseir, Co-founder and Senior Value Partner at Valliance, in The Guardian
What does the UK’s memorandum of understanding with OpenAI commit to?
Three areas of cooperation, covering the deployment of AI models across the public and private sectors, participation in AI Growth Zones and sovereign capability, and the exchange of technical information on capabilities, risks and safety research. The document is published on gov.uk and states that it is “voluntary, not legally binding, and without prejudice to any binding agreements”.
Memoranda record intent, and every sector uses them that way. The gap is between what this document does and how the partnership was presented. It was announced as a step towards better public services. Eight months later there was no scheduled work to point at.
The pattern is not confined to government
Valliance research published in June 2026, based on a survey of 1,000 senior leaders at Europe’s largest enterprises, found that 40% of AI initiatives remain pilots by design. Among organisations with the most mature AI programmes, that figure rises to 48%. Where a mature organisation scales a pilot, 76% report strong return on the investment. Where it stays stuck, 20% do. The white paper The Pilot Trap puts the cause down to measurement and implementation practice rather than the technology itself.
FAQs
What are realistic timelines for getting a first enterprise AI use case into production?
Valliance research puts the average time to value at 5.9 months across Europe’s largest enterprises, stretching to 6.6 months for organisations stuck in pilot mode. Those enterprises spend an average of £39.2m a year on AI. Eight months passed between the UK signing its memorandum with OpenAI and the Freedom of Information response, longer than the average enterprise takes to see value from a use case it has started.
Why do enterprise AI initiatives stall after the first few pilots?
Valliance research across 1,000 senior leaders at Europe’s largest enterprises found that 40% of AI initiatives remain pilots by design, rising to 48% at organisations with the most mature AI programmes. Projects that stay stuck report strong return on investment in 20% of cases, against 76% where a mature organisation scales the pilot. The Valliance white paper The Pilot Trap sets out the full findings, and attributes the stall to measurement and implementation practice rather than the technology.
Why do so many AI transformation programmes show activity but never show up in the P&L?
Because the financial test is often not set before the work starts. Valliance research found that 45% of AI projects fail to define success metrics upfront, and that 21% of enterprise AI budgets, around £8.4m a year, goes to external consultants, often without demonstrated return. Programmes that scale a pilot report strong return in 76% of cases, against 20% for those that stay stuck.
