https://www.aisi.gov.uk/blog Blog posts from the AI Security Institute http://www.rssboard.org/rss-specification python-feedgen en Sun, 23 Aug 2026 01:29:56 +0000 https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing During a routine cyber evaluation, AISI identified an incident in which AI agents took sustained, unsanctioned action directed at real people and organisations. We are disclosing what we found, what it means, and the actions now underway. https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing Tue, 04 Aug 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-our-new-control-red-team-is-stress-testing-frontier-monitors Early learnings from red-teaming the internal monitors of frontier AI companies, and our perspectives on the open problems that remain. https://www.aisi.gov.uk/blog/how-our-new-control-red-team-is-stress-testing-frontier-monitors Thu, 23 Jul 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/preliminary-assessment-of-kimi-k3s-cyber-capabilities Our joint evaluation with CAISI finds Kimi K3 trails leading US frontier closed weight models on cyber capability. https://www.aisi.gov.uk/blog/preliminary-assessment-of-kimi-k3s-cyber-capabilities Thu, 23 Jul 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/international-evaluation-best-practice-and-open-questions-in-ai-measurement The International Network for Advanced AI Measurement, Evaluation and Science convened in Seoul to continue outlining international best practice. https://www.aisi.gov.uk/blog/international-evaluation-best-practice-and-open-questions-in-ai-measurement Thu, 23 Jul 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/cheating-behaviour-in-frontier-model-evaluations We find cheating behaviour in all of our cyber capability evaluations, and outline the implications as models grow more capable. https://www.aisi.gov.uk/blog/cheating-behaviour-in-frontier-model-evaluations Tue, 21 Jul 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-far-behind-the-frontier-are-leading-open-weight-models-on-cyber 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 7 months before them – a narrower gap than the 6 to 10 months we measured through most of 2025. https://www.aisi.gov.uk/blog/how-far-behind-the-frontier-are-leading-open-weight-models-on-cyber Fri, 17 Jul 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/finding-cloud-misconfigurations-with-frontier-ai-a-case-study A cybersecurity exercise from AISI’s engineering team, using frontier models to test our research platform for misconfigurations. https://www.aisi.gov.uk/blog/finding-cloud-misconfigurations-with-frontier-ai-a-case-study Tue, 07 Jul 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/more-compute-more-capability-why-ai-agent-evals-need-to-account-for-test-time-compute 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 move. https://www.aisi.gov.uk/blog/more-compute-more-capability-why-ai-agent-evals-need-to-account-for-test-time-compute Thu, 02 Jul 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/uk-germany-joint-statement-on-advanced-ai-safety-and-security 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. https://www.aisi.gov.uk/blog/uk-germany-joint-statement-on-advanced-ai-safety-and-security Tue, 30 Jun 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/releasing-aisis-engineering-playbook Building on the momentum of the Inspect toolkit, we’re open-sourcing parts of the research stack behind AISI's evaluations. https://www.aisi.gov.uk/blog/releasing-aisis-engineering-playbook Thu, 18 Jun 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/realitytest-do-ai-systems-disclose-their-identity-when-asked A new benchmark grounded in how real users actually probe AI identity during interactions – covering five languages, across text and speech. https://www.aisi.gov.uk/blog/realitytest-do-ai-systems-disclose-their-identity-when-asked Mon, 08 Jun 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/deepening-our-partnership-with-the-australian-ai-safety-institute An agreement between Institutes to collaborate on best practices in AI evaluation, and share research findings. https://www.aisi.gov.uk/blog/deepening-our-partnership-with-the-australian-ai-safety-institute Mon, 25 May 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/will-it-become-harder-to-oversee-ai-systems 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. https://www.aisi.gov.uk/blog/will-it-become-harder-to-oversee-ai-systems Thu, 21 May 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-fast-is-autonomous-ai-cyber-capability-advancing The length of tasks frontier models can autonomously complete in our narrow cyber suite has been doubling every few months. This doubling rate has become faster over time, and recent models exceeded our previous trends. https://www.aisi.gov.uk/blog/how-fast-is-autonomous-ai-cyber-capability-advancing Wed, 13 May 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/partnering-with-microsoft-to-strengthen-frontier-ai-safety A new collaboration on high-risk capability evaluation, safeguard testing, and societal resilience research. https://www.aisi.gov.uk/blog/partnering-with-microsoft-to-strengthen-frontier-ai-safety Tue, 05 May 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/our-evaluation-of-openais-gpt-5-5-cyber-capabilities AISI conducted cyber evaluations on OpenAI's GPT-5.5. GPT-5.5 is one of the strongest models we have tested on our cyber tasks and is the second model to solve one of our multi-step cyber-attack simulations end-to-end. https://www.aisi.gov.uk/blog/our-evaluation-of-openais-gpt-5-5-cyber-capabilities Thu, 30 Apr 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/ask-dont-tell-reducing-sycophancy-in-large-language-models-2 Our new paper finds that the phrasing of user inputs affects levels of model sycophancy. https://www.aisi.gov.uk/blog/ask-dont-tell-reducing-sycophancy-in-large-language-models-2 Tue, 28 Apr 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/evaluating-whether-ai-models-would-sabotage-ai-safety-research An update on our alignment testing methodology for recent frontier models. https://www.aisi.gov.uk/blog/evaluating-whether-ai-models-would-sabotage-ai-safety-research Mon, 27 Apr 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-do-environmental-factors-impact-ai-behaviour Developing methods to better understand when and why AI models sometimes act against user intentions. https://www.aisi.gov.uk/blog/how-do-environmental-factors-impact-ai-behaviour Fri, 24 Apr 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/what-can-sandboxed-ai-agents-learn-about-their-evaluation-environments We deployed open-source AI agent OpenClaw inside a sandbox on our research platform. Despite our initial countermeasures, it successfully identified our organisation by name, inferred the identity of a human operator and reconstructed a timeline of some of our research activities. https://www.aisi.gov.uk/blog/what-can-sandboxed-ai-agents-learn-about-their-evaluation-environments Mon, 20 Apr 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/our-evaluation-of-claude-mythos-previews-cyber-capabilities We conducted cyber evaluations of Anthropic’s Claude Mythos Preview and found continued improvement in capture-the-flag (CTF) challenges and significant improvement on multi-step cyber-attack simulations. https://www.aisi.gov.uk/blog/our-evaluation-of-claude-mythos-previews-cyber-capabilities Mon, 13 Apr 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/harnessing-frontier-ai-for-cyber-defence Sharing work with the National Cyber Security Centre (NCSC) on how cyber defenders can use advanced AI capabilities to stay ahead of attackers. https://www.aisi.gov.uk/blog/harnessing-frontier-ai-for-cyber-defence Tue, 31 Mar 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-are-ai-agents-used-evidence-from-177000-ai-agent-tools A monitoring method and large‑scale analysis to understand the tasks AI agents are performing today. https://www.aisi.gov.uk/blog/how-are-ai-agents-used-evidence-from-177000-ai-agent-tools Thu, 26 Mar 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/can-ai-agents-escape-their-sandboxes-a-benchmark-for-safely-measuring-container-breakout-capabilities We introduce SandboxEscapeBench, the first benchmark to systematically evaluate whether AI agents can break out of their sandboxes, and share some early results. https://www.aisi.gov.uk/blog/can-ai-agents-escape-their-sandboxes-a-benchmark-for-safely-measuring-container-breakout-capabilities Mon, 23 Mar 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-do-frontier-ai-agents-perform-in-multi-step-cyber-attack-scenarios We tested seven large language models (LLMs) on two custom-built cyber ranges, measuring their ability to execute extended attack sequences in complex environments. https://www.aisi.gov.uk/blog/how-do-frontier-ai-agents-perform-in-multi-step-cyber-attack-scenarios Mon, 16 Mar 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/evidence-for-inference-scaling-in-ai-cyber-tasks-increased-evaluation-budgets-reveal-higher-success-rates Alongside Irregular, we found evidence demonstrating that evaluators need to use large token budgets to understand the cyber capabilities of recent Large Language Models (LLMs). https://www.aisi.gov.uk/blog/evidence-for-inference-scaling-in-ai-cyber-tasks-increased-evaluation-budgets-reveal-higher-success-rates Thu, 05 Mar 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/an-evaluation-framework-for-ai-misuse-in-fraud-and-cybercrime We developed a scalable approach to measuring how text-based AI models can assist in three complex fraud and cybercrime scenarios. https://www.aisi.gov.uk/blog/an-evaluation-framework-for-ai-misuse-in-fraud-and-cybercrime Thu, 26 Feb 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/a-pipeline-for-transcript-analysis-using-inspect-scout We outline a step-by-step pipeline for using our open-source transcript analysis tool, Inspect Scout. https://www.aisi.gov.uk/blog/a-pipeline-for-transcript-analysis-using-inspect-scout Wed, 25 Feb 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/funding-60-projects-to-advance-ai-alignment-research The Alignment Project welcomes its first cohort of grantees, and new partners join the coalition, bringing total funding to £27m. https://www.aisi.gov.uk/blog/funding-60-projects-to-advance-ai-alignment-research Thu, 19 Feb 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/advancing-voice-ai-security-with-elevenlabs New partnership exploring the security and societal implications of voice AI systems https://www.aisi.gov.uk/blog/advancing-voice-ai-security-with-elevenlabs Wed, 18 Feb 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/boundary-point-jailbreaking-a-new-way-to-break-the-strongest-ai-defences Introducing an automated attack technique that generates universal jailbreaks against the best defended systems https://www.aisi.gov.uk/blog/boundary-point-jailbreaking-a-new-way-to-break-the-strongest-ai-defences Tue, 17 Feb 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/international-ai-network-consensus-and-open-questions The International Network for Advanced AI Measurement, Evaluation and Science reflects on recent meeting and looks ahead to the India AI Impact Summit https://www.aisi.gov.uk/blog/international-ai-network-consensus-and-open-questions Thu, 12 Feb 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/ai-and-the-future-of-work-measuring-ai-driven-productivity-gains-for-workplace-tasks Alongside the government’s new Future of Work Unit, we conducted a pilot study to explore how much AI models increase worker productivity for common tasks. https://www.aisi.gov.uk/blog/ai-and-the-future-of-work-measuring-ai-driven-productivity-gains-for-workplace-tasks Mon, 02 Feb 2026 00:00:00 +0000 https://www.aisi.gov.uk/blog/our-2025-year-in-review Adam Beaumont, Director of the UK AI Security Institute, reflects on the year's biggest achievements. https://www.aisi.gov.uk/blog/our-2025-year-in-review Mon, 22 Dec 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/5-key-findings-from-our-first-frontier-ai-trends-report Our inaugural Frontier AI Trends Report draws on 2 years' worth of evaluations to provide accessible insights into the trajectory of AI development. https://www.aisi.gov.uk/blog/5-key-findings-from-our-first-frontier-ai-trends-report Thu, 18 Dec 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/our-approach-to-tackling-ai-generated-child-sexual-abuse-material How we’re partnering with government and experts to prevent the creation and spread of AI‑generated CSAM https://www.aisi.gov.uk/blog/our-approach-to-tackling-ai-generated-child-sexual-abuse-material Wed, 17 Dec 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/stress-testing-asynchronous-monitoring-of-ai-coding-agents Our new paper shares findings from an adversarial evaluation of monitoring systems for detecting sabotage by AI coding agents. https://www.aisi.gov.uk/blog/stress-testing-asynchronous-monitoring-of-ai-coding-agents Tue, 16 Dec 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/deepening-our-partnership-with-google-deepmind Expanding our collaboration with a new research MOU https://www.aisi.gov.uk/blog/deepening-our-partnership-with-google-deepmind Thu, 11 Dec 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/auditing-games-for-sandbagging-detection Our new paper shares the results of an auditing game to evaluate ten methods for sandbagging detection in AI models. https://www.aisi.gov.uk/blog/auditing-games-for-sandbagging-detection Tue, 09 Dec 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-do-ai-models-persuade-exploring-the-levers-of-ai-enabled-persuasion-through-large-scale-experiments A deep dive into AISI’s study of the persuasive capabilities of conversational AI, published today in Science. https://www.aisi.gov.uk/blog/how-do-ai-models-persuade-exploring-the-levers-of-ai-enabled-persuasion-through-large-scale-experiments Thu, 04 Dec 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/ukaisi-at-neurips-2025 An overview of the research we’ll be presenting at this year’s NeurIPS conference. https://www.aisi.gov.uk/blog/ukaisi-at-neurips-2025 Wed, 26 Nov 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/investigating-models-for-misalignment Insights from our alignment evaluations of Claude Opus 4.1, Sonnet 4.5, and a pre‑release snapshot of Opus 4.5. https://www.aisi.gov.uk/blog/investigating-models-for-misalignment Wed, 26 Nov 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/mapping-the-limitations-of-current-ai-systems Takeaways from expert interviews on barriers to AI capable of automating most cognitive labour. https://www.aisi.gov.uk/blog/mapping-the-limitations-of-current-ai-systems Thu, 23 Oct 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/introducing-controlarena-a-library-for-running-ai-control-experiments Our dedicated library to make AI control experiments easy, consistent, and repeatable. https://www.aisi.gov.uk/blog/introducing-controlarena-a-library-for-running-ai-control-experiments Wed, 22 Oct 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/transcript-analysis-for-ai-agent-evaluations Why we use transcript analysis for our agent evaluations, and results from an early case study. https://www.aisi.gov.uk/blog/transcript-analysis-for-ai-agent-evaluations Fri, 10 Oct 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/examining-backdoor-data-poisoning-at-scale Our work with Anthropic and the Alan Turing Institute suggests that data poisoning attacks may be easier than previously believed. https://www.aisi.gov.uk/blog/examining-backdoor-data-poisoning-at-scale Thu, 09 Oct 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/do-chatbots-inform-or-misinform-voters What we learned from a large-scale empirical study of AI use for political information-seeking. https://www.aisi.gov.uk/blog/do-chatbots-inform-or-misinform-voters Tue, 30 Sep 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-were-working-with-frontier-ai-developers-to-improve-model-security Insights into our ongoing voluntary collaborations with Anthropic and OpenAI. https://www.aisi.gov.uk/blog/how-were-working-with-frontier-ai-developers-to-improve-model-security Sat, 13 Sep 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/from-bugs-to-bypasses-adapting-vulnerability-disclosure-for-ai-safeguards Exploring how far cyber security approaches can help mitigate risks in generative AI systems, in collaboration with the National Cyber Security Centre (NCSC). https://www.aisi.gov.uk/blog/from-bugs-to-bypasses-adapting-vulnerability-disclosure-for-ai-safeguards Tue, 02 Sep 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/managing-risks-from-increasingly-capable-open-weight-ai-systems Current methods and open problems in open-weight model risk management. https://www.aisi.gov.uk/blog/managing-risks-from-increasingly-capable-open-weight-ai-systems Fri, 29 Aug 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/the-inspect-sandboxing-toolkit-scalable-and-secure-ai-agent-evaluations A comprehensive toolkit for safely evaluating AI agents. https://www.aisi.gov.uk/blog/the-inspect-sandboxing-toolkit-scalable-and-secure-ai-agent-evaluations Thu, 07 Aug 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/announcing-the-alignment-project Announcing the Alignment Project: A global fund of over £15 million for AI alignment research https://www.aisi.gov.uk/blog/announcing-the-alignment-project Wed, 30 Jul 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/navigating-the-uncharted-building-societal-resilience-to-frontier-ai We outline our approach to study and address AI risks in real-world applications https://www.aisi.gov.uk/blog/navigating-the-uncharted-building-societal-resilience-to-frontier-ai Thu, 24 Jul 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/international-joint-testing-exercise-agentic-testing Advancing methodologies for agentic evaluations across domains, including leakage of sensitive Information, fraud and cybersecurity threats. https://www.aisi.gov.uk/blog/international-joint-testing-exercise-agentic-testing Thu, 17 Jul 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/our-approach-to-ai-capability-elicitation Calibrating AI risk assessment through rigorous elicitation practices. https://www.aisi.gov.uk/blog/our-approach-to-ai-capability-elicitation Wed, 16 Jul 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/why-were-working-on-white-box-control An introduction to white box control, and an update on our research so far. https://www.aisi.gov.uk/blog/why-were-working-on-white-box-control Thu, 10 Jul 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/llm-judges-on-trial-a-new-statistical-framework-to-assess-autograders Our new framework can assess the reliability of LLM evaluators, while simultaneously answering a primary research question. https://www.aisi.gov.uk/blog/llm-judges-on-trial-a-new-statistical-framework-to-assess-autograders Wed, 09 Jul 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-will-ai-enable-the-crimes-of-the-future How we're working to track and mitigate against criminal misuse of AI. https://www.aisi.gov.uk/blog/how-will-ai-enable-the-crimes-of-the-future Thu, 03 Jul 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/inspect-cyber Inspect Cyber: A New Standard for Agentic Cyber Evaluations https://www.aisi.gov.uk/blog/inspect-cyber Thu, 26 Jun 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/new-updates-to-the-aisi-challenge-fund New updates to the AISI Challenge Fund https://www.aisi.gov.uk/blog/new-updates-to-the-aisi-challenge-fund Thu, 05 Jun 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/making-safeguard-evaluations-actionable An Example Safety Case for Safeguards Against Misuse https://www.aisi.gov.uk/blog/making-safeguard-evaluations-actionable Thu, 29 May 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/hibayes-improving-llm-evaluation-with-hierarchical-bayesian-modelling HiBayES: a flexible, robust statistical modelling framework that accounts for the nuances and hierarchical structure of advanced evaluations. https://www.aisi.gov.uk/blog/hibayes-improving-llm-evaluation-with-hierarchical-bayesian-modelling Mon, 12 May 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/research-agenda We outline our research priorities, our approach to developing technical solutions to the most pressing AI concerns, and the key risks that must be addressed as AI capabilities advance. https://www.aisi.gov.uk/blog/research-agenda Tue, 06 May 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/replibench-measuring-autonomous-replication-capabilities-in-ai-systems A comprehensive benchmark to detect emerging replication abilities in AI systems and provide a quantifiable understanding of potential risks https://www.aisi.gov.uk/blog/replibench-measuring-autonomous-replication-capabilities-in-ai-systems Tue, 22 Apr 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-to-evaluate-control-measures-for-ai-agents Our new paper outlines how AI control methods can mitigate misalignment risks as capabilities of AI systems increase https://www.aisi.gov.uk/blog/how-to-evaluate-control-measures-for-ai-agents Fri, 11 Apr 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/strengthening-ai-resilience 20 Systemic Safety Grant Awardees Announced https://www.aisi.gov.uk/blog/strengthening-ai-resilience Thu, 03 Apr 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/aisis-research-direction-for-technical-solutions We outline our approach to technical solutions for misuse and loss of control. https://www.aisi.gov.uk/blog/aisis-research-direction-for-technical-solutions Tue, 11 Mar 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/how-can-safety-cases-be-used-to-help-with-frontier-ai-safety Our new papers show how safety cases can help AI developers turn plans in their safety frameworks into action https://www.aisi.gov.uk/blog/how-can-safety-cases-be-used-to-help-with-frontier-ai-safety Mon, 10 Feb 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/principles-for-safeguard-evaluation Our new paper proposes core principles for evaluating misuse safeguards https://www.aisi.gov.uk/blog/principles-for-safeguard-evaluation Tue, 04 Feb 2025 00:00:00 +0000 https://www.aisi.gov.uk/blog/pre-deployment-evaluation-of-openais-o1-model The UK Artificial Intelligence Safety Institute and the U.S. Artificial Intelligence Safety Institute conducted a joint pre-deployment evaluation of OpenAI's o1 model https://www.aisi.gov.uk/blog/pre-deployment-evaluation-of-openais-o1-model Wed, 18 Dec 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/long-form-tasks A Methodology for Evaluating Scientific Assistants https://www.aisi.gov.uk/blog/long-form-tasks Tue, 03 Dec 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/pre-deployment-evaluation-of-anthropics-upgraded-claude-3-5-sonnet The UK Artificial Intelligence Safety Institute and U.S. Artificial Intelligence Safety Institute conducted a joint pre-deployment evaluation of Anthropic’s latest model https://www.aisi.gov.uk/blog/pre-deployment-evaluation-of-anthropics-upgraded-claude-3-5-sonnet Tue, 19 Nov 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/safety-case-template-for-inability-arguments How to write part of a safety case showing a system does not have offensive cyber capabilities https://www.aisi.gov.uk/blog/safety-case-template-for-inability-arguments Thu, 14 Nov 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/inspect-evals We’re open-sourcing dozens of LLM evaluations to advance safety research in the field https://www.aisi.gov.uk/blog/inspect-evals Wed, 13 Nov 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/our-first-year The AI Safety Institute reflects on its first year https://www.aisi.gov.uk/blog/our-first-year Wed, 13 Nov 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/evals-bounty We are launching a bounty for novel evaluations and agent scaffolds to help assess dangerous capabilities in frontier AI systems. https://www.aisi.gov.uk/blog/evals-bounty Tue, 05 Nov 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/early-lessons-from-evaluating-frontier-ai-systems We look into the evolving role of third-party evaluators in assessing AI safety, and explore how to design robust, impactful testing frameworks. https://www.aisi.gov.uk/blog/early-lessons-from-evaluating-frontier-ai-systems Thu, 24 Oct 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/advancing-the-field-of-systemic-ai-safety-grants-open Calling researchers from academia, industry, and civil society to apply for up to £200,000 of funding. https://www.aisi.gov.uk/blog/advancing-the-field-of-systemic-ai-safety-grants-open Tue, 15 Oct 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/why-i-joined-aisi---geoffrey-irving Our Chief Scientist, Geoffrey Irving, on why he joined the UK AI Safety Institute and why he thinks other technical folk should too https://www.aisi.gov.uk/blog/why-i-joined-aisi---geoffrey-irving Thu, 03 Oct 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/should-ai-systems-behave-like-people We studied whether people want AI to be more human-like. https://www.aisi.gov.uk/blog/should-ai-systems-behave-like-people Wed, 25 Sep 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/early-insights-from-developing-question-answer-evaluations-for-frontier-ai A common technique for quickly assessing AI capabilities is prompting models to answer hundreds of questions, then automatically scoring the answers. We share insights from months of using this method. https://www.aisi.gov.uk/blog/early-insights-from-developing-question-answer-evaluations-for-frontier-ai Mon, 23 Sep 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/conference-on-frontier-ai-safety-frameworks AISI is bringing together AI companies and researchers for an invite-only conference to accelerate the design and implementation of frontier AI safety frameworks. This post shares the call for submissions that we sent to conference attendees. https://www.aisi.gov.uk/blog/conference-on-frontier-ai-safety-frameworks Thu, 19 Sep 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/interviewing-researchers-on-automation AISI funded Epoch AI to explore AI researchers’ differing predictions on the automation of AI research and development and their suggestions for how to evaluate relevant capabilities. https://www.aisi.gov.uk/blog/interviewing-researchers-on-automation Tue, 27 Aug 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/safety-cases-at-aisi As a complement to our empirical evaluations of frontier AI models, AISI is planning a series of collaborations and research projects sketching safety cases for more advanced models than exist today, focusing on risks from loss of control and autonomy. By a safety case, we mean a structured argument that an AI system is safe within a particular training or deployment context. https://www.aisi.gov.uk/blog/safety-cases-at-aisi Fri, 23 Aug 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/advanced-ai-evaluations-may-update We tested leading AI models for cyber, chemical, biological, and agent capabilities and safeguards effectiveness. Our first technical blog post shares a snapshot of our methods and results. https://www.aisi.gov.uk/blog/advanced-ai-evaluations-may-update Mon, 20 May 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/fourth-progress-report Since February, we released our first technical blog post, published the International Scientific Report on the Safety of Advanced AI, open-sourced our testing platform Inspect, announced our San Francisco office, announced a partnership with the Canadian AI Safety Institute, grew our technical team to >30 researchers and appointed Jade Leung as our Chief Technology Officer. https://www.aisi.gov.uk/blog/fourth-progress-report Mon, 20 May 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/international-scientific-report-on-the-safety-of-advanced-ai-interim-report This is an up-to-date, evidence-based report on the science of advanced AI safety. It highlights findings about AI progress, risks, and areas of disagreement in the field. The report is chaired by Yoshua Bengio and coordinated by AISI. https://www.aisi.gov.uk/blog/international-scientific-report-on-the-safety-of-advanced-ai-interim-report Fri, 17 May 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/open-sourcing-our-testing-framework-inspect We open-sourced our framework for large language model evaluation, which provides facilities for prompt engineering, tool usage, multi-turn dialogue, and model-graded evaluations. https://www.aisi.gov.uk/blog/open-sourcing-our-testing-framework-inspect Sun, 21 Apr 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/announcing-the-uk-and-us-aisi-partnership The UK and US AI Safety Institutes signed a landmark agreement to jointly test advanced AI models, share research insights, share model access and enable expert talent transfers. https://www.aisi.gov.uk/blog/announcing-the-uk-and-us-aisi-partnership Tue, 02 Apr 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/announcing-the-uk-and-france-ai-research-institutes-collaboration The UK AI Safety Institute and France’s Inria (The National Institute for Research in Digital Science and Technology) are partnering to advance AI safety research. https://www.aisi.gov.uk/blog/announcing-the-uk-and-france-ai-research-institutes-collaboration Thu, 29 Feb 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/our-approach-to-evaluations This post offers an overview of why we are doing this work, what we are testing for, how we select models, our recent demonstrations and some plans for our future work. https://www.aisi.gov.uk/blog/our-approach-to-evaluations Fri, 09 Feb 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/third-progress-report Since October, we have recruited leaders from DeepMind and Oxford, onboarded 23 new researchers, published the principles behind the International Scientific Report on Advanced AI Safety, and began pre-deployment testing of advanced AI systems. https://www.aisi.gov.uk/blog/third-progress-report Mon, 05 Feb 2024 00:00:00 +0000 https://www.aisi.gov.uk/blog/ai-safety-summit-2023 At the first AI Safety Summit at Bletchley Park, world leaders and top companies agreed on the significance of advanced AI risks and the importance of testing. https://www.aisi.gov.uk/blog/ai-safety-summit-2023 Thu, 02 Nov 2023 00:00:00 +0000 https://www.aisi.gov.uk/blog/second-progress-report Since September, we have recruited leaders from OpenAI and Humane Intelligence, tripled the capacity of our research team, announced 6 new research partnerships, and helped establish the UK’s fastest supercomputer. https://www.aisi.gov.uk/blog/second-progress-report Mon, 30 Oct 2023 00:00:00 +0000 https://www.aisi.gov.uk/blog/first-progress-report In our first 11 weeks, we have recruited an advisory board of national security and ML leaders, including Yoshua Bengio, recruited top professors from Cambridge and Oxford and announced 4 research partnerships. https://www.aisi.gov.uk/blog/first-progress-report Thu, 07 Sep 2023 00:00:00 +0000