AISI Research & Publications | The AI Security Institute
https://www.aisi.gov.uk/research
UK AI Safety Institute research — named as an external evaluator in the OpenAI system card, and one of the few public bodies running frontier evaluations
20
Lines added
0
Lines removed
Was read
2026-10-08
8630b1b8e147a48c…
Changed by
2026-10-27
9729b7298b885aef…
20 lines added
Left column is the version read 2026-10-08, right is 2026-10-27. Unchanged stretches are elided. Both bodies are kept in full and are addressed by the hashes above.
Showing the first 400 lines of the comparison; 279 further lines are held back. This page reports a change, it is not a copy of anyone’s document — the source is linked above and is the authority for what it says.
⋯ 18 unchanged lines
1919 Research Agenda
2020
2121 Publications and blogs:
2222
23+The Effectiveness and Durability of Pretraining Data Filtering in Protein Language Models
24+
25+Chem-Bio
26+
27+•
28+Oct 26, 2026
29+Research
30+
31+Read research
32+
33+ReproBAIT: Can coding agents build their own biological AI tools?
34+
35+Chem-Bio
36+
37+•
38+Oct 26, 2026
39+Research
40+
41+Read research
42+
2343 Transect: Retaining Observability for Long-Horizon LLM Agent Evaluations
2444
2545 Science of Evaluations
2646