News
News follow-ups
Short dated pieces on stories the earlier site reported: what the primary record shows now, with every source and the date it was read.
- The US-China talks on AI: what happened sinceOne meeting in Geneva in May 2024, a leaders' statement on nuclear weapons in November 2024, and in September 2026 a new U.S.-China Super Intelligence Dialogue with its first exchange due by November 2026.Published 29 October 2026 · Follows a post of 14 May 2024
- The EU AI Act: what happened sinceIn force since 1 August 2024; the Digital Omnibus on AI of July 2026 moved the high-risk rules to 2 December 2027 (Annex III) and 2 August 2028 (Annex I).Published 29 October 2026 · Follows a post of 10 June 2024
- Frontier model safety: what happened sinceGovernment testing agreements in August 2024, a renamed federal centre, and a June 2026 executive order asking for up to 30 days of voluntary pre-release access to the most capable models.Published 29 October 2026 · Follows a post of 28 May 2024
- AI self-regulation: what happened sinceCalifornia made frontier developers' safety frameworks a legal duty in September 2025; a December 2025 executive order set up a task force to challenge state AI laws; one company's framework reached version 3.4 in July 2026.Published 29 October 2026 · Follows a post of 26 May 2024
- Local government AI savings: what happened since at DerbyA £12.25 million a year target set in December 2023; by May 2026 the council's assistants had handled more than 3.2 million enquiries, and the target is still the only savings figure it has published.Published 29 October 2026 · Follows a post of 3 June 2024 · Data: CSV
- The energy use of AI: what happened sinceU.S. data centres used 4.4% of the country's electricity in 2023 and could use 6.7% to 12% by 2028, in the report the Department of Energy published in December 2024.Published 29 October 2026 · Follows a post of 6 April 2024 · Data: CSV
- The Stanford AI Index: what happened sinceThe 2026 edition counts 362 documented AI incidents in 2025, up from 233, and 150 AI-related bills passed by U.S. state legislatures in 2025.Published 29 October 2026 · Follows a post of 19 April 2024 · Data: CSV
- Frontier model risk ratings: what happened sinceGPT-4o's "medium" of August 2024 came from a four-category scorecard; the April 2025 framework dropped persuasion, and GPT-5 was treated as High in biological and chemical capability.Published 29 October 2026 · Follows a post of 15 August 2024
- Canada's $2.4 billion for AI: what happened since$2 billion of the April 2024 package was for computing; it became a three-part compute strategy, and in 2026 a national strategy with targets to 2034.Published 29 October 2026 · Follows a post of 9 April 2024 · Data: CSV