Sean O hEigeartaigh

Based on 2021/22 Census data, 16% of the overall UK population was born overseas, so 12.5% of prisoners being born overseas means foreign-born people are under-represented in the prison population. But hatred of foreigners sells more newspapers, unfortunately.

Based on 2021/22 Census data, 16% of the overall UK population was born overseas, so 12.5% of prisoners being born overseas means foreign-born people are under-represented in the prison population. But hatred of foreigners sells, unfortunately.

Obituary: Professor Margaret Boden - LCFI The Leverhulme Centre for the Future of Intelligence (CFI) community is mourning the loss of Professor Margaret Boden, OBE ScD FBA, who recently passed away. Professor Boden was Research Professor of ...

Sad to hear of the passing of the great Margaret Boden. She gave a few wonderful lectures in the early days of our two centres in Cambridge. Towering intellectual contributions in cognitive science, AI, philosophy and psychology,plus a wry sense of humour. www.lcfi.ac.uk/news-events/...

Lots of highlights from Shanghai, one was the release of Shanghai AI Lab's Frontier AI Safety framework, co-developed w Concordia AI. Will have significant influence in Shanghai++ ecosystem.
concordia-ai.com/research/fro...

Exciting week ahead with the World AI Conference in Shanghai and a series of AI safety & governance workshops alongside. I'll be offline - be good and don't make AI recursively self-improve, etc.

I was delighted to be an advisor to this excellent work by Liam Epstein - exploring the idea of sovereign wealth funds for managing the transition to transformative AI.
www.convergenceanalysis.org/fellowships/...

Developing responsible AI: Vision and practice Amidst accelerating advancements in AI and growing disparities in digital access, the World Internet Conference (WIC) is going to convene a multidisciplinary coalition—including members of WIC Special...

Off to Geneva for the ITU AI for Good conference. Tomorrow I'll be speaking at a workshop the World Internet Conference is organising with ITU. My talk:
"AI Governance is in Crisis".

Because, well, I think AI governance is in crisis.
aiforgood.itu.int/event/develo...

How the striking of a GOP regulatory ban will affect the global artificial intelligence race The House just voted to pass the One Big Beautiful Bill without a moratorium that would ban states from enforcing AI regulations for 10 years.

Pleased to have my work quoted in this excellent @thebulletin.org article on the push against AI regulation in the US, the spectre of the race with China, and the false claim made by tech companies incl OpenAI, and US politicians, that China isn't regulating AI.
thebulletin.org/2025/07/how-...

The Singapore Consensus on Global AI Safety Research Priorities Rapidly improving AI capabilities and autonomy hold significant promise of transformation, but are also driving vigorous debate on how to ensure that AI is safe, i.e., trustworthy, reliable, and secur...

Delighted to have contributed in a small way to the Singapore Consensus - an important step forward at a critical time in establishing shared, international agreement on the safety challenges we can address together. My deepest thanks to everyone who made it happen.
arxiv.org/abs/2506.20702

In research on global risk, work on how catastrophes occur and spread in complex interconnected systems (systemic risk, cascading risk, polycrisis) has rarely been connected with work on global catastrophic risk (worst-case outcomes). Our new paper bridges this gap: www.cambridge.org/core/journal...

In our scramble to win the AI race against China, we risk losing ourselves A fraud is being perpetuated on the American people and our pliant, gullible political leaders.

Good post from US Senator Chris Murphy.
"The leaders of the artificial intelligence industry in the United States... claim that any meaningful regulation of AI in America will allow China to leapfrog the United States. But they are dead wrong." www.chrismurphyct.com/p/in-our-scr...

Risk Tiers: Towards a Gold Standard for Advanced AI - Oxford Martin AIGI Increasing risks from advanced AI demand effective risk management systems tailored to this rapidly changing technology. One key part of risk management is establishing risk tiers. Risk tiers are cate...

Pleased to have contributed to this research memo on AI risk tiers, led by the excellent Nick Caputo and Oxford Martin AI Governance Initiative:
aigi.ox.ac.uk/publications...

Celebrating small milestones: 10 years at Cambridge earlier this year. Can't believe they're still letting me do this for a living. For all the many ways academia sucks, nothing beats a small town full of interesting people absolutely, obsessively passionate about what they do.

How Some of China’s Top AI Thinkers Built Their Own AI Safety Institute The emergence of the China AI Safety and Development Association (CnAISDA) is a pivotal moment for China’s frontier AI governance. How it navigates substantial domestic challenges and growing geopolit...

Nice paper by Scott Singer, Karson Elmgren & Oliver Guest on the evolution of China's AISI.

As Singer notes, this is being led by experts who are influential in Chinese policy, and genuinely care about frontier AI safety and the need for cooperation.
carnegieendowment.org/research/202...

Field moving so quickly you need to put up preprints - fine, understood. But these then get picked up by media as settled science, even as the scientific community is doing 'live' peer review and analysing approach & community. Critiques/further analysis rarely make media.

Advanced AI suffers ‘complete accuracy collapse’ in face of complex problems, study finds ‘Pretty devastating’ Apple paper raises doubts about race to reach stage of AI at which it matches human intelligence

Worth knowing this paper's conclusions have been brought into question by IMO well-founded critiques (I've shared some). AFAIK original paper not formally peer-reviewed. Illustrates a problem in field (why I emphasised my recent paper pre-peer review)
www.theguardian.com/technology/2...

Opinion | Anthropic C.E.O.: Don’t Let A.I. Companies off the Hook

And again: REALLY pleased to see (excellent) articles like this from Dario Amodei.
www.nytimes.com/2025/06/05/o...

... is it the weather?

Pending review process, this will hopefully be out later this year in an exciting Springer Nature book on the future of AI, being led by Lord Fairfax and Max Rangeley. Huge thanks to colleagues who gave feedback on earlier drafts. Thoughts & discussion as always appreciated! 5/5

My views in this paper are shaped by analysis of a great many documents and speeches, but also many trips to both US and China over the last few years (I was in China 4x last year)! 4/5

It's not too late though: there are steps we can take: policymakers, academics, researchers. Above all, I believe it is time to challenge this most dangerous of fictions. 3/5

The evidence is weak, & the narrative is being mostly promoted in the West, and often by actors who stand to benefit directly from it. It strangles prospects for international cooperation when we most need it, & creates some of the most dangerous conditions for pursuit of AGI.2/5

New working paper (pre-review), maybe my most important in recent years. I examine the evidence for the US-China race to AGI and decisive strategic advantage, & analyse the impact this narrative is having on our prospects for cooperation on safety. 1/5
papers.ssrn.com/sol3/papers....

But we love you! Can't we persuade you to stay?

(for avoidance of doubt given this is the Internet after all, this is me gently poking fun at the Giant Boogeyman narratives around EA on both the left and the right, rather than endorsing said narratives)

The reason for being on both Bluesky and Twitter is that it's very important to know that the effective altruists both engineered the rise of the far right AND are responsible for Woke AI, leftist globalism and the evil Let's Regulate AI Influence Operation.

And another good role: 1 year postdoc position on AI/bio risk:
www.jobs.cam.ac.uk/job/51468/

Great speech from Singapore's President on the need for AI regulation.

"We can't leave it to the future to see how much bad actually comes out of the AI race."
www.istana.gov.sg/Newsroom/Spe...

Behind the Curtain: Top AI CEO foresees white-collar bloodbath Hardly anyone is paying attention.

"Amodei said AI companies and government need to stop "sugar-coating" what's coming: the possible mass elimination of jobs across technology, finance, law, consulting and other white-collar professions, especially entry-level gigs."
www.axios.com/2025/05/28/a...

Important topic, and a good opportunity for the right person: research assistant role on AI/nuclear risk
www.jobs.cam.ac.uk/job/51403/