Sapiens
Policy

What are AI safety institutes?

Published June 1, 2026 · 5 min read

AI SAFETY INSTITUTESInspected before it carries passengers.A model is tested first — then the gate opens to the public.AI safetyinstituteTestResearchShareAIthe modelgate opensusers & businessesLike aviation inspectors checking a new aircraft, they test a model before it reaches the public.

Definition

An AI safety institute is a government-backed organization that tests and researches the most advanced (“frontier”) AI models to find and reduce serious risks.

At a glance

  • State-backed bodies (not private firms) with three jobs: test frontier models, do safety research, share findings.
  • The UK and US launched the first two at the November 2023 UK AI Safety Summit.
  • An 11-member International Network (US, UK, EU, Japan, France, Canada, Singapore, South Korea, Australia, Kenya) coordinates them[2].
  • Both flagships rebranded toward security in 2025: UK to AI Security Institute, US to CAISI[5].

What they do

They run technical tests (evaluations) on the most powerful AI to check for dangerous capabilities, such as aiding cyberattacks, bio/chemical weapons, or systems acting on their own[1]. They also publish safety research and guidance.

Why it matters for your business

Their standards are becoming the benchmark for “responsible” AI, shaping regulation and what you should ask AI vendors. The US body (now CAISI) is industry’s main government contact for AI testing[4]. Consistent cross-border standards, aligned through the network and a US-UK partnership, mean fewer conflicting national rules[3].

Bottom line

They are the public sector’s inspectors for frontier AI, and the source of the testing standards your future regulators and vendors will rely on.

References

  1. Artificial intelligence safety institute. Wikipedia en.wikipedia.org
  2. FACT SHEET: U.S. Department of Commerce & U.S. Department of State Launch the International Network of AI Safety Institutes — U.S. Department of Commerce, NIST. NIST www.nist.gov
  3. U.S. and UK Announce Partnership on Science of AI Safety. U.S. Department of Commerce www.commerce.gov
  4. Statement from U.S. Secretary of Commerce Howard Lutnick on Transforming the U.S. AI Safety Institute into the U.S. Center for AI Standards and Innovation — Howard Lutnick. U.S. Department of Commerce www.commerce.gov
  5. Inside the U.K.'s Bold Experiment in AI Safety. TIME time.com

Comments

Questions, corrections, and links welcome. Be specific and civil.

  • Loading comments…