# UK AI Safety Institute Mission The UK AI Safety Institute (AISI), established in November 2023, is a state-backed body tasked with evaluating frontier AI models to ensure their safety. It was renamed the AI Security Institute in 2025 and collaborates with an international network of similar institutes. The UK AI Safety Institute (AISI) is a state-backed organization established in November 2023 to evaluate and ensure the safety of advanced artificial intelligence (AI) models, often referred to as frontier AI models. It originated from the Frontier AI Taskforce and was officially launched during the AI Safety Summit hosted by the United Kingdom. The institute's primary mission is to conduct independent safety evaluations of cutting-edge AI systems, addressing concerns that AI companies cannot reliably "mark their own homework" in assessing risks. In 2025, the institute was renamed the AI Security Institute, reflecting a broader focus on security-related aspects of AI safety. The creation of the UK AISI was part of a global movement that gained momentum in 2023, driven by public declarations about potential existential risks from AI. The institute operates from its headquarters in London and opened an office in San Francisco in May 2024 to be closer to major AI developers. It collaborates with international counterparts through a network of AI safety institutes, aiming to set new, international standards for AI safety evaluation. ## Establishment and Early Development The UK AISI was formally established in November 2023 during the AI Safety Summit, an event convened by then-Prime Minister Rishi Sunak. Sunak had expressed his intention to "make the UK not just the intellectual home but the geographical home of global AI safety regulation," emphasizing the need for independent oversight. The institute evolved from the Frontier AI Taskforce, a government body created earlier in 2023 to assess risks associated with advanced AI. Its initial focus was on evaluating large language models and other generative AI systems for potential harms, including bias, misinformation, and misuse. In April 2024, the UK AISI concluded an agreement with the US AI Safety Institute to collaborate on at least one joint safety test. This partnership aimed to harmonize evaluation procedures and reduce duplication of efforts. Politico reported in April 2024 that many AI companies had not shared pre-deployment access to their most advanced models for evaluation, partly because they were waiting for the UK and US institutes to agree on common rules. Nick Clegg, president of global affairs at Meta, noted that companies sought clarity on evaluation standards before granting access. The institute announced in May 2024 that it would open an office in San Francisco, where many leading AI companies, such as [OpenAI](https://www.wikiprompt.org/wiki/openai) and [Anthropic](https://www.wikiprompt.org/wiki/anthropic), are based. This move was part of a plan to "set new, international standards on AI safety," according to UK technology minister Michele Donelan. The San Francisco office was intended to facilitate direct engagement with developers and enable pre-deployment testing of frontier models. ## Mission and Objectives The core mission of the UK AISI is to evaluate frontier AI models to identify potential safety risks before they are deployed. This includes assessing capabilities in areas such as cybersecurity, biosecurity, and autonomous decision-making. The institute aims to provide evidence-based guidance to policymakers and developers, helping to mitigate risks associated with advanced AI. Its work is grounded in the principle of independent evaluation, ensuring that safety assessments are not influenced by commercial interests. The institute also contributes to the development of international standards for AI safety. By collaborating with other national institutes, it seeks to create a common framework for evaluating AI systems, which can be adopted globally. This includes sharing methodologies, conducting joint tests, and publishing findings to inform public discourse. The institute's research often involves stress-testing models from major developers, including [Google DeepMind](https://www.wikiprompt.org/wiki/google-deepmind) and other leading labs. In 2025, the institute was renamed the AI Security Institute, signaling an expanded focus on security-related threats, such as the use of AI in cyberattacks or the proliferation of harmful content. This change reflected evolving priorities in AI governance, as concerns shifted from general safety to specific security implications. ## International Network At the AI Seoul Summit in May 2024, international leaders agreed to form a network of AI safety institutes, comprising bodies from the UK, the US, Japan, France, Germany, Italy, Singapore, South Korea, Australia, Canada, and the European Union. The UK AISI plays a central role in this network, which aims to coordinate evaluation efforts and share best practices. The network held an exercise in July 2025 to explore issues with evaluating AI agents, particularly regarding sensitive information leakage and cybersecurity. Members also met at NeurIPS 2025 in San Diego to discuss recent advances in AI safety research. The network's formation marked a significant step toward global cooperation on AI governance. Each member institute operates within its national context but contributes to a collective effort to ensure the safe development of frontier AI. The UK AISI's collaboration with the US institute, in particular, has been instrumental in aligning evaluation standards across the two leading AI nations. ## Specific Institutes ### Australia The Australian AI Safety Institute was announced on 25 November 2025 by the Albanese government. It is housed within the Department of Industry, Science and Resources and is supported by a budget of A$29,900,000 over four years. Its general manager is Kate Conroy, who also leads responsible AI initiatives in the Royal Australian Air Force. The institute focuses on evaluating AI systems for domestic applications, including defense and public services. ### Canada Canada announced plans to create an AI safety institute in April 2024, and the institute was officially founded in November 2024. It operates under Innovation, Science and Economic Development Canada, in partnership with the Canadian Institute for Advanced Research (CIFAR). The institute is supported by a budget of CA$50,000,000 over five years, funding research and evaluation activities across various sectors. ### European Union The EU AI Office, founded in May 2024, is a member of the international network of AI safety institutes. It coordinates AI safety efforts across member states, focusing on regulatory compliance and risk assessment. The office works closely with national institutes to ensure a cohesive approach to AI governance within the EU. ### France France created the Institut national pour l'évaluation et la sécurité de l'intelligence artificielle (INESIA) on 31 January 2025. This national institute for AI evaluation and security is tasked with assessing the safety of AI systems used in French industries and public services. It collaborates with other European institutes to harmonize evaluation methodologies. ### India The Ministry of Electronics and Information Technology (MeitY) held consultations on 7 October 2024 with various stakeholders, including Meta Platforms, Google, Microsoft, IBM, [OpenAI](https://www.wikiprompt.org/wiki/openai), NASSCOM, and Indian Institutes of Technology (IITs), regarding the establishment of an AI Safety Institute. The decision was made to shift focus from regulation to standards-setting, risk identification, and damage detection. The institute may use the ₹20 crore allotted to the Safe and Trusted Pillar of the IndiaAI Mission for its initial budget. Minister Ashwini Vaishnaw announced the creation of the IndiaAI Safety Institute on 30 January 2025, to promote domestic R&D grounded in India's social, economic, cultural, and linguistic diversity. It operates under a "hub-and-spoke" model with collaboration from academic institutions, tech firms, and international organizations like UNESCO. ### Japan The Japan AISI (J-AISI) was founded in February 2024 as part of the Information Technology Promotion Agency. It employs about 23 people and consists of the Council of AISI, the AISI Steering Committee, and a secretariat with six teams. Akiko Murakami, previously of IBM Japan and Sompo Japan, serves as executive director, with Kenji Hiramoto and Suguru Nishimura as deputy executive directors. The institute focuses on evaluating AI systems for Japanese market needs. ### Kenya Kenya agreed to join the international network of AI safety institutes but has not announced specific details. It is the only African state in the network, highlighting the growing global reach of AI safety efforts. ### Singapore The Digital Trust Centre, initially founded in June 2022, was renamed the Singapore AISI in May 2024. Part of Nanyang Technological University, the institute partners with the Infocomm Media Development Authority and is supported by an investment of S$10,000,000 per year. It focuses on evaluating AI systems for trust and safety in Southeast Asia. ### South Korea South Korea announced in May 2024 that it would create an AI safety institute under the Electronics and Telecommunications Research Institute. It is supported by a tentative investment of 10 to 20 million South Korean won per year and employs at least 30 people. The institute was founded in November 2024 and is based in Bundang District. ## Impact and Future Directions The UK AISI has significantly influenced global AI safety discourse, setting precedents for independent evaluation. Its work has informed policy decisions in the UK and abroad, contributing to the development of regulatory frameworks. The institute's collaboration with international partners has facilitated the sharing of evaluation techniques and risk assessment tools. Looking ahead, the institute faces challenges in keeping pace with rapid advancements in AI, particularly in areas like [generative AI](https://www.wikiprompt.org/wiki/generative-ai) and autonomous systems. Its renamed focus on security suggests a shift toward addressing immediate threats, such as AI-enabled cyberattacks. The institute's role in the international network will likely expand as more countries establish their own safety bodies. The UK AISI's mission remains central to ensuring that frontier AI models are developed and deployed responsibly. By providing rigorous, independent evaluations, it aims to mitigate risks while fostering innovation. As AI technologies evolve, the institute's methods and priorities will need to adapt, but its core objective of safeguarding public interests endures. --- Source: https://www.wikiprompt.org/wiki/uk-ai-safety-institute-mission License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/) Last updated: 2026-09-12T16:23:41.102569+00:00