Verified, then logged.
Sun 22 Nov 2026 · 00:21 KST / 15:21 UTC한국어 | ENSearch /

Breaking04 / Policy

AISI selects 60 projects for AI alignment funding

OpenAI and Microsoft joined as new partners as total support for the research programme rose to £27m.

Two researchers seen from behind compare an AI model’s reasoning with an independent verification on adjacent monitors in a naturally lit research lab. AI-generated image

(Global) The UK's AI Security Institute (AISI) announced funding for 60 AI alignment research projects, including work by LawZero. It said the next application round would open in summer.

(Global) AISI's announcement describes AI alignment as the challenge of developing AI that follows human intent and remains under human oversight and control. It cautioned that methods used for today's models may cease to work as systems become more capable and autonomous.

(Global) AISI announced £12m in additional funding, bringing total support to £27m. OpenAI and Microsoft joined the programme as new partners, with OpenAI contributing £5.6m.

(Global) AISI said it selected projects after assessing criteria including relevance, feasibility and innovation. The selected research spans mathematics, learning theory, economics and cognitive science.

(Global) AISI described LawZero's Scientist AI as a project focused on assessing information reliability and making model reasoning easier to inspect. Its design separates a component that proposes hypotheses from another that checks them.

(Global) AISI has not disclosed the exact application dates for the next round, which it said would open in summer.