UK BUSINESS AI UNIVERSE: Safety

Every record on this page, generated from data.json. Click a name to open its panel at the top of the page. Records resting on one source are tagged "one source".

Record	Type	First source	Confidence	ID

What this page covers

As written in the research. Data checked 29 October 2026.

UK companies and non-profits that test, audit, secure and govern AI systems, research how to make advanced AI safe, and detect deepfakes.

In scope are UK-headquartered (or UK-incorporated) companies and non-profits whose main business is AI evaluation, red-teaming, AI and agent security, AI audit and assurance, AI governance software, alignment and interpretability research, or deepfake detection; Apollo Research is counted as UK by Alex's decision. Out of scope are government bodies (the AI Security Institute is on the Government site), general cyber security firms that use AI but do not secure AI (CybSafe, Senseon), and Conjecture, which is on the Labs page.

Kept in the data, not shown

2Records the research marked unverified stay in the data files and never render (decision 4.3 #7).

BIZ-SAFE-011 Geordie AI (cross-listing)
BIZ-SAFE-012 Trent AI (cross-listing)

Downloads

Open data for this page. Sterling conversions: Bank of England daily spot rate (series XUDLUSS, XUDLERS).

safety-entries.csv data.json deals.json events.json news.json