Overview
- Liz Kendall used a Royal United Services Institute speech to argue that Britain should build AI sovereignty by backing UK firms and working with fellow “middle powers” to reduce over‑reliance on a few dominant players.
- The government will publish an AI Hardware Plan in June to secure capability across the chip stack, building on £100 million through ARIA for novel AI hardware and a further £100 million to buy semiconductors for public data centers.
- The Sovereign AI Unit now holds £500 million to invest through direct support and procurement, with a linked fund offering £1 million to £9 million grants aimed first at high‑value datasets and automated labs for faster R&D.
- The UK’s AI Security Institute said it independently tested Anthropic’s restricted Mythos model and will share evaluation best practice at a July meeting of global AI safety bodies, a push to export UK know‑how on model testing.
- Kendall framed the approach as pro‑alliance and pro‑investment with the U.S., signaled the UK will not align with the EU AI Act, noted a U.S. Tech Prosperity Deal remains frozen, and floated a 5% share of the global AI‑chip market as an ambition while the hardware plan’s details are still being written.