Item Response Theory for AI Safety
Applies Item Response Theory to eight safety benchmarks across 192 models, finding few latent factors, cheap adaptive evaluation subsets, and detection of sandbagging and silent API model changes.
Research / Sources
UK AI Security Institute
The UK government’s AI Security Institute: evaluations, red-teaming and research on frontier AI risks. aisi.gov.uk
Organisation (36)Red Team (13)Science of Evaluations (11)Cyber & Autonomous Systems (10)Engineering (6)Human Influence (5)Control (4)Societal Resilience (4)Safety Cases (3)Chem-Bio (2)Model Transparency (2)Strategic Awareness (1)
Focus All
Topic Trustworthy AI
Date Any time
Advancing AI voice security with ElevenLabs
UK AISI announces a research partnership with ElevenLabs on voice AI security, studying how people distinguish AI from humans in real-time voice and how perceived identity affects trust. No results yet.
Do chatbots inform or misinform voters?
UK study of voters found about one in eight used AI chatbots for political information, and chatbot users became no less informed than search users.
Conversational AI increases political knowledge as effectively as self-directed internet search
Survey and randomised trials find 13% of UK voters used chatbots for political information, and chatbot conversations raised political knowledge as much as internet search without increasing misinformation belief.
The levers of political persuasion with conversational AI
Across 76,977 participants and 19 LLMs, shows post-training and prompting boost political persuasiveness most, and that persuasive gains coincide with lower factual accuracy.
Should AI systems behave like people?
Surveys public preferences on whether AI should behave like people, motivated by risks of anthropomorphic AI such as impersonation, deception and emotional over-attachment.