By Roman Aglaganov
In November 2023, the British government invited world leaders and tech executives to Bletchley Park, the country house where Alan Turing and his colleagues broke Nazi codes. The symbolism was not subtle: the birthplace of modern computing, hosting a conversation about computing’s next big risk.
Bletchley, 2023: the risks
The first AI Safety Summit produced the Bletchley Declaration, signed by 28 countries and the European Union, including both the United States and China, a rare joint signature. It recognised that the most capable “frontier” AI could cause serious, even catastrophic, harm, and called for international cooperation on safety research. The UK also launched its AI Safety Institute to test advanced models, renamed the AI Security Institute in 2025.
Seoul, 2024: the commitments
Six months later in Seoul, sixteen AI companies signed voluntary Frontier AI Safety Commitments, promising to publish how they assess and limit the risks of their most powerful models.
Paris, 2025: the opportunities
By February 2025 the tone had changed. The Paris AI Action Summit, co-chaired by France and India, focused on investment, jobs and public-interest AI, and drew more than a thousand participants from over 100 countries. The US vice-president warned against excessive regulation, and both the United States and the United Kingdom declined to sign the summit’s final declaration.
What it tells us: governments agree AI matters; they disagree on whether the priority is guard rails or growth. The next summit, hosted by India in February 2026, moved the conversation further towards impact on development. Bletchley’s ghosts, one suspects, would have had views.
