Communiqués de presse AI Security Institute (UK) — Flux RSS et veille en direct
PPN World indexe les communiqués de presse de AI Security Institute (UK) et en archive chaque publication. Ce fil est silencieux depuis le 23 juillet 2026 — ses 50 communiqués indexés restent consultables, et une alerte vous préviendra dès qu'il publiera de nouveau.
URL du flux RSS de AI Security Institute (UK)
https://hpampthnhbmgpdxezfop.supabase.co/functions/v1/scrape-config?config=aisi-uk-blogÀ propos de ce fil
AI Security Institute (UK) fait partie des 1200+ fils de presse que PPN World surveille en continu — un fil en anglais axé sur le secteur Intelligence artificielle, classé sous « Royaume-Uni » dans notre index par pays. Chaque communiqué est dédupliqué, géolocalisé et enrichi d'un résumé IA, et vous pouvez y attacher des alertes sur recherches enregistrées. Le communiqué le plus récent de notre index a été publié le 23 juillet 2026; le fil est silencieux depuis.
Comment suivre les communiqués de AI Security Institute (UK)
- Copiez l'URL du flux RSS de AI Security Institute (UK) affichée ci-dessus.
- Collez-la dans n'importe quel lecteur RSS (Feedly, Inoreader, NetNewsWire…) pour recevoir le flux brut.
- Ou suivez AI Security Institute (UK) dans PPN World pour ajouter déduplication, alertes sur recherches enregistrées et résumés IA sur les 1200+ fils.
Communiqués de AI Security Institute (UK) — archives
- How our new Control Red Team is stress-testing frontier monitorsil y a 7 j
Early learnings from red-teaming the internal monitors of frontier AI companies, and our perspectives on the open problems that remain.
- UK AISI / CAISI Preliminary Assessment of Kimi K3's Cyber Capabilitiesil y a 7 j
Our joint evaluation with CAISI finds Kimi K3 trails leading US frontier closed weight models on cyber capability.
- International evaluation best practice and open questions in AI measurementil y a 7 j
The International Network for Advanced AI Measurement, Evaluation and Science convened in Seoul to continue outlining international best practice.
- Cheating behaviour in frontier model evaluationsil y a 9 j
We find cheating behaviour in all of our cyber capability evaluations, and outline the implications as models grow more capable.
- How Far Behind the Frontier are Leading Open Weight Models on Cyber?il y a 13 j
We evaluated the cyber capabilities of leading open and closed weight AI models, and found that recent open models GLM-5.2 and DeepSeek V4-Pro perform similarly to frontier closed models released 4 to…
- Finding Cloud Misconfigurations with Frontier AI: A Case Studyil y a 23 j
A cybersecurity exercise from AISI’s engineering team, using frontier models to test our research platform for misconfigurations.
- More compute, more capability: Why AI agent evaluations need to account for test-time computeil y a 28 j
Standard evaluations cap how much compute AI agents can use. We show that raising those caps changes measured capability, the difficulty of tasks agents can solve, and how fast the frontier appears to…
- UK-Germany Joint Statement on advanced AI safety and securityil y a 30 j
A joint statement by the UK and Germany on collaborating to ensure advanced AI is developed safely and its risks are rigorously understood and managed.
- Releasing AISI’s Engineering Playbookil y a 42 j
Building on the momentum of the Inspect toolkit, we’re open-sourcing parts of the research stack behind AISI's evaluations.
- RealityTest: Do AI systems disclose their identity when asked?il y a 52 j
A new benchmark grounded in how real users actually probe AI identity during interactions – covering five languages, across text and speech.
- Deepening our partnership with the Australian AI Safety Instituteil y a 66 j
An agreement between Institutes to collaborate on best practices in AI evaluation, and share research findings.
- Will it become harder to oversee AI systems?il y a 70 j
Our new report examines today’s AI oversight landscape, how robust it is to capability advances, and the pathways that could lead to its degradation.
Les plus récents communiqués de ce fil défilent en direct pour les membres PPN World.