AI Security Institute
United Kingdom · Governance body · UK Government
Our summary
The AI Security Institute is a government-backed organization established to advance safe, secure and beneficial advanced AI. It conducts technical research to understand AI capabilities and risks, evaluates threats to national security and public safety, and develops risk mitigation approaches. The institute also works with AI developers, policymakers and allied governments to inform responsible AI development and global policy.
Written by the Index in our own words. It is not legal advice; the official text is the authority.
Inside this policy
Every section of this policy in order. Our notes summarise each one in plain language and do not quote the official text, which is the authority.
Drafted by the Index An automated summary of the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
10 sections
Jump to a chapter
Rigorous AI research to enable advanced AI governance
1 section
This section introduces the organisation's focus on ensuring advanced AI systems are safe, secure and beneficial through rigorous research.
Governments have a critical role to play in ensuring advanced AI is safe, secure and beneficial
The AI Security Institute is a government-backed organisation that conducts research and develops infrastructure to assess advanced AI capabilities and impacts, while collaborating with researchers, developers and other governments to influence AI development and global policy.
Featured work
3 sections
This section lists selected research projects and partnerships, including studies on open-weight model capabilities, AI persuasion, and collaboration with Google DeepMind.
How Far Behind the Frontier are Leading Open Weight Models on Cyber?
The Institute evaluated cyber capabilities of leading AI models and found that recent open-weight models perform comparably to frontier closed models with only a 4 to 7 month delay, narrower than the historical 6 to 10 month gap.
How do AI models persuade? Exploring the levers of AI-enabled persuasion through large-scale experiments
The Institute published research in Science examining how conversational AI systems can persuade through large-scale experimental evaluation.
Deepening our partnership with Google DeepMind
The Institute expanded its research collaboration with Google DeepMind through a new memorandum of understanding.
Our mission is to equip governments with a scientific understanding of the risks posed by advanced AI
2 sections
This section outlines the Institute's mission to provide governments with scientific evidence on advanced AI risks, covering technical research and global impact activities.
Technical research
The Institute monitors AI development trends, evaluates risks to national security and public safety, and develops technical solutions including safeguards, alignment and control mechanisms.
Global impact
The Institute engages with AI developers on responsible development, advises policymakers on current and emerging AI risks, and shares findings with allied governments.
Join us to shape the trajectory of AI
The Institute seeks top talent for its mission and operates with over 100 technical staff drawn from leading research organisations, offering substantial funding, computing resources, priority model access and partnerships with leading research institutions.
Controls
What this policy requires, forbids, grants or penalises, one control at a time and in our own words. Each control links to the official text, which is the authority. This is not legal advice.
Drafted by the Index Drafted automatically from the official text, not yet reviewed by a person.Reviewed Checked by a person against the official text.
Themes covered 8 of 35
Choose a theme to see only its controls. About the themes
Risk and scope 2 of 4
- Prohibited practices, not covered
- Impact assessment, not covered
Data and privacy 0 of 4
- Data governance and quality, not covered
- Data subject rights and consent, not covered
- Cross-border data transfer, not covered
- Copyright and training data, not covered
Transparency and records 1 of 4
- Labelling of synthetic content, not covered
- Technical documentation, not covered
- Record keeping and logging, not covered
Technical safeguards 2 of 4
- Bias and non-discrimination, not covered
- Testing and evaluation, not covered
Assurance and lifecycle 0 of 4
- Conformity assessment and certification, not covered
- Registration and notification, not covered
- Incident and breach reporting, not covered
- Post-market monitoring and corrective action, not covered
Rights and protected groups 0 of 4
- Automated decision-making rights, not covered
- Children and vulnerable groups, not covered
- Biometric identification and surveillance, not covered
- Whistleblowing, complaints and redress, not covered
Roles and organisation 3 of 6
- General purpose and foundation models, not covered
- Public sector use and procurement, not covered
- Innovation support and sandboxes, not covered
Enforcement and sectors 0 of 5
- Enforcement, penalties and regulators, not covered
- Sector rules: health, not covered
- Sector rules: finance and insurance, not covered
- Sector rules: employment, not covered
- Sector rules: education, not covered
The full list of controls could not be loaded, so only the first 6 are shown.
Governments have a critical role to play in ensuring advanced AI is safe, secure and beneficial 1 control
- ObligationApplies to: Public authorityDrafted by the Index
A government-backed organisation conducts research and develops infrastructure to assess advanced AI capabilities and impacts, and collaborates with researchers, developers and other governments to influence AI development and global policy.
Technical research 1 control
- ObligationApplies to: Public authorityDrafted by the Index
The Institute should monitor trends in AI development, assess risks to national security and public safety, and develop technical solutions for safeguards, alignment and control.
Global impact 3 controls
- ObligationApplies to: Public authorityDrafted by the Index
The Institute should work with AI developers to promote responsible development practices.
- ObligationApplies to: Public authorityDrafted by the Index
The Institute should inform policymakers about current and emerging AI risks.
- ObligationApplies to: Public authorityDrafted by the Index
The Institute should collaborate and share its findings with allied governments.
Join us to shape the trajectory of AI 1 control
- ObligationApplies to: Public authorityDrafted by the Index
The Institute commits to operating with substantial funding, computing resources, priority access to top models and partnerships with leading research organisations to support its mission.
Change history
No changes recorded since we began monitoring this policy.