TL;DR:
- Sir Keir Starmer’s former chief of staff Morgan McSweeney held talks with Google DeepMind after leaving Downing Street in February, pitching a project at the intersection of AI and democratic politics.
- He met DeepMind’s head of frontier policy and public affairs Owen Larter and approached three or four tech companies in total, according to people briefed on the discussions.
- DeepMind says it is “not working with Mr McSweeney on any projects”; allies say there are “no live discussions” and he has moved on to other ventures.
Morgan McSweeney, the Labour strategist widely credited with engineering Sir Keir Starmer’s 2024 landslide, pitched Google DeepMind on an AI project shortly after resigning as the Prime Minister’s chief of staff in February, according to the Financial Times.
The pitch and the participants
McSweeney approached three or four technology companies about a venture focused on what one person described as “the gap opening up between government and AI capability”. Meetings included Owen Larter, DeepMind’s head of frontier policy and public affairs. McSweeney is also said to have been interested in how political parties could use AI to understand voters and shape electoral campaigns — a direct evolution of the data-driven targeting that characterised his work for Starmer. A spokesperson for DeepMind said the company is not working with him on any projects, and one ally said there were “no live discussions” and that he had moved on.
The story reflects a pattern already visible in Washington: former senior political operatives gravitating towards frontier AI labs as a next step, combining campaign-era data instincts with the labs’ policy-engagement teams. DeepMind’s London base and UK Government access make it a natural venue for anyone thinking through AI in state delivery and electoral politics simultaneously.
Political context
McSweeney resigned in February saying he took “full responsibility” for advising Starmer to appoint Lord Peter Mandelson as UK ambassador to Washington, after Mandelson’s prior links became untenable. Within Labour, he is credited with driving the party rightward on welfare, migration and leftwing factions to blunt the threat from Reform UK, a strategy that has lost the party ground with voters drifting to the Greens on the left. His post-Downing Street pivot to AI suggests he sees the next political frontier as algorithmic — both in how governments deliver services and how campaigns reach voters.
Looking Forward
The immediate story is a near-miss rather than a deal, but the direction of travel matters for UK AI governance. Anyone mapping the intersection of DeepMind’s London policy apparatus, DSIT priorities under Liz Kendall and Dan Jarvis, and campaign-data infrastructure will be doing so from the outside for now. That is a meaningful gap. Expect McSweeney’s project interests — AI in state delivery, electoral-data applications — to surface through whichever firm or venture he eventually joins.