UK AI Security Institute
To minimize surprise to the UK and humanity from rapid and unexpected advances in AI by conducting rigorous research and model evaluations.
At a Glance
- AI Model Developers
- Government Agencies
- Research Institutions
- International Regulatory Bodies
AI Tools by UK AI Security Institute
(1)
Inspect AI
Open Source LLM Eval Framework
Discussions
No discussions yet
Be the first to start a discussion about UK AI Security Institute
Latest News
AISI launches call for information on secure AI infrastructure.
UK AI Security Institute releases inaugural Frontier AI Trends Report.
AI Safety Institute rebranded as AI Security Institute at Paris Summit.
AISI signs historic model access agreements with OpenAI, Anthropic, and Google DeepMind.
Products & Services
An open-source software library/framework for testing and evaluating the safety and capabilities of LLMs.
A periodic research publication analyzing the performance and risk trends of the latest frontier AI models.
Technical testing of advanced AI models for risks related to cybersecurity, biological misuse, and autonomous behavior.
Market Position
First state-backed organization dedicated to advanced AI safety; positioned as a global leader in technical model evaluations and research.
Leadership
Founders
Ian Hogarth
Chair of AISI; Co-founder of Songkick and venture capitalist (Plural); appointed to lead the Frontier AI Taskforce in 2023.
Rishi Sunak
Former UK Prime Minister who spearheaded the creation of the Frontier AI Taskforce and the AI Safety Institute at Bletchley Park.
Executive Team
Adam Beaumont
Interim Director
Former Chief AI Officer at GCHQ (UK Intelligence Agency).
Ian Hogarth
Chair
Tech entrepreneur and investor; former founder of Songkick.
Board of Directors
Founding Story
Originally established in April 2023 as the 'Frontier AI Taskforce' to advise the UK government on the risks of advanced AI models. It was formally launched as the AI Safety Institute (AISI) at the AI Safety Summit in November 2023 at Bletchley Park. In February 2025, it was rebranded as the AI Security Institute to focus more on national security risks and technical evaluations.
Business Model
Revenue Model
Publicly funded government agency; non-profit mission focused on public safety and national security.
Pricing Tiers
All research reports and the Inspect framework are open-source and free to use.
Target Markets
- AI Model Developers
- Government Agencies
- Research Institutions
- International Regulatory Bodies
- Pre-deployment safety testing for AI labs
- Government risk assessment for national security
- Informing global AI safety standards
- Advancing scientific understanding of model capabilities
- OpenAI
- Google DeepMind
- Anthropic
- Mistral AI