This page is still being polished. If you have thoughts, please share them via the feedback form.
Data on this page is preliminary and may change. Please do not share or cite these figures publicly.
Technical mechanisms and engineering interventions that directly modify how an AI system processes inputs, generates outputs, or operates, including changes to models, training procedures, runtime behaviors, and supporting hardware.
AI systems that are trustworthy, reliable and secure by design give people the confidence to embrace and adopt AI innovation. Following a classic safety engineering framework, the research areas in this category involves: A. Specifying and validating the desired behaviour – This includes technical methods to address the complex challenges in specifying system behaviours in a way that accurately captures the desired intent without causing undesired side effects, for both singlestakeholder settings (e.g. reward hacking, scalable methods to discover specification loopholes) and multi-stakeholder settings (e.g. balancing competing preferences, ethical and legal alignment). B. Designing a system that meets the specification – This covers techniques for training models – both closed and open weights – that are trustworthy (e.g. reducing confabulation, increasing robustness against tampering), alternative finetuning methods to make specific precise changes to an AI system (e.g. model editing), and methods to build AI systems in a way that are guaranteed to meet their specifications (e.g. verifiable programme synthesis, world models with formal guarantees). C. Verifying that the AI system meets its specification – This entails techniques to provide high-confidence assurances that an AI system meets its specifications (e.g. formal verification), including in novel contexts (e.g. robustness testing), as well as interpretability techniques to look into the black box to understand why the AI system behaves the way it does (e.g. mechanistic interpretability).
Reasoning
Foundational research into specification, design, and verification techniques for safer AI system development.
Specification & Validation: Defining the system’s purpose
Specification involves defining desired system behaviour, whereas validation ensures that the specification meets the needs of the user, developer, or society – did I build the right system? In other words, specification and validation require confronting the complexity of defining objectives in a way that captures user or societal intent without omitting important constraints or causing undesired side effects, as well as dealing with disagreement and tradeoffs between diverse stakeholders.
2.2.2 Testing & EvaluationDesign and implementation: Building the system
This section focuses on techniques to make systems that meet their specifications. The design and implementation process involves sourcing data, pretraining models, post-training models, and integrating them into AI systems.
2.4 Engineering & DevelopmentVerification: Assessing if the system works as specified
The research areas described in this section aim to assess the extent to which the built system (2.2) meets its specifications (2.1). This section discusses several broad types of techniques that can be used to provide evidence that a system is safe. In practice, the effectiveness of these methods is often limited by access and transparency, but they can play a central role in constructing AI safety cases: structured arguments for why systems pose an acceptably low level of risk (Clymer, Buhl).
2.2 Risk & AssuranceRisk Assessment
The primary goal of risk assessment is to understand the severity and likelihood of a potential harm. Risk assessments are used to prioritise risks and determine if they cross thresholds that demand specific action. Consequential development and deployment decisions are predicated on these assessments. The research areas in this category involve: A. Developing methods to measure the impact of AI systems for both current and future AI – This includes developing standardised assessments for risky behaviours of AI systems through audit techniques and benchmarks, evaluation and assessment of new capabilities, including potentially dangerous ones; and for real-world societal impact such as labour, misinformation and privacy through field tests and prospective risks analysis. B. Enhancing metrology to ensure that the measurements are precise and repeatable – This includes research in technical methods for quantitative risk assessment tailored to AI systems to reduce uncertainty and the need for large safety margins. This is an important open area of research. C. Building enablers for third-party audits to support independent validation of risk assessments – This includes developing secure infrastructure that enables thorough evaluation while protecting intellectual property, including preventing model theft.
2.2.1 Risk AssessmentRisk Assessment > Audit techniques and benchmarks
Techniques and benchmarks with which AI systems can be effectively and efficiently tested for harmful behaviours are highly varied and central to risk assessments (IAISR, Birhane-A).
3.2.1 Benchmarks & EvaluationRisk Assessment > Downstream impact assessment and forecasting
Assessing and forecasting the many societal impacts of AI systems is one of the most central goals of risk assessments.
2.2.1 Risk AssessmentRisk Assessment > Secure evaluation infrastructure
External auditors and oversight bodies need infrastructure and protocols that enable thorough evaluation while protecting sensitive intellectual property. Ideally, evaluation infrastructure should enable double-blindness: the evaluator’s inability to directly access the system’s parameters and developers’ inability to know what exact evaluations are run (Reuel, Bucknall-A, Casper-B). Meanwhile, the importance of mutual security will continue to grow as system capabilities and risks increase. Methods for developing secure infrastructure for auditing and oversight are known to be possible.
3.2.2 Technical StandardsRisk Assessment > System safety assessment
Safety assessment is not just about individual AI systems, but also their interaction with the rest of the world. For example, when an AI company discovers concerning behaviour from their system, the resulting risks depend, in part, on having internal processes in place to escalate the issue to senior leadership and work to mitigate the risks. System safety considers both AI systems and the broader context that they are deployed in. The study of system safety focuses on the interactions between different technical components as well as processes and incentives in an organisation (IAISR, Hendrycks-B, AISES, Alaga).
2.2.1 Risk AssessmentRisk Assessment > Metrology for AI risk assessment
Metrology, the science of measurement, has only recently been studied in the context of AI risk assessment (IAISR, Hobbhahn). Current approaches generally lack standardisation, repeatability, and precision.
3.2.1 Benchmarks & EvaluationThe Singapore Consensus on Global AI Safety Research Priorities
Bengio, Yoshua; Maharaj, Tegan; Ong, C.-H. Luke; Russell, Stuart D.; Song, Dawn; Tegmark, Max; Lan, Xue; Zhang, Ya-Qin; Casper, Stephen; Lee, Wan Sie; Mindermann, Sören; Wilfred, Vanessa; Balachandran, Vidhisha; Barez, Fazl; Belinsky, Michael; Bello, Imane; Bourgon, Malo; Brakel, Mark; Campos, Siméon; Cass-Beggs, Duncan; Chen, Jiahao; Chowdhury, Rumman; Seah, Kuan Chua; Clune, Jeff; Dai, Jie; Delaborde, Agnes; Dziri, Nouha; Eiras, Francisco; Engels, Joshua; Fan, Jinyu; Gleave, Adam; Goodman, Noah D.; Heide, Fynn; Heidecke, Johannes; Hendrycks, Dan; Hodes, Cyrus; Hsiang, Bryan Low Kian; Huang, Minlie; Jawhar, Sami; Wang, Jingyu; Kalai, Adam Tauman; Kamphuis, Meindert; Kankanhalli, Mohan; Kantamneni, Subhash; Kirk, M.; Kwa, Thomas; Ladish, Jeffrey; Lam, Kwok-Yan; Lee, Wan Sie; Lee, Taewhi; Li, Xiaopeng; Liu, Jiajun; Lu, Ching-Cheng; Mai, Yifan; Mallah, Richard; Michael, Julian; Moës, Nick; Møller, Simon Geir; Nam, K. H.; Ng, TP; Nitzberg, Mark; Nushi, Besmira; Ó hÉigeartaigh, Seán; Ortega, Alejandro; Peigné, Pierre; Petrie, J. Howard; Prud'homme, Benjamin; Rabbany, Reihaneh; Sanchez-Pi, Nayat; Schwettmann, Sarah; Shlegeris, Buck; Siddiqui, Saad; Sinha, Ashish; Soto, Martín; Tan, Cheston; Dong, Ting; Tjhi, William; Trager, Robert; Tse, Brian; Tung, Anthony K. H.; Willes, John; Wong, David; Xu, Wei; Xu, Rong; Zeng, Yi; Zhang, Hao; Žikelić, Djordje (2025)
This is the first International AI Safety Report. Following an interim publication in May 2024, a diverse group of 96 Artificial Intelligence (AI) experts contributed to this first full report, including an international Expert Advisory Panel nominated by 30 countries, the Organisation for Economic Co-operation and Development (OECD), the European Union (EU), and the United Nations (UN). The report aims to provide scientific information that will support informed policymaking. It does not recommend specific policies…. This report summarises the scientific evidence on the safety of general-purpose AI. The purpose of this report is to help create a shared international understanding of risks from advanced AI and how they can be mitigated. To achieve this, this report focuses on general-purpose AI – or AI that can perform a wide variety of tasks – since this type of AI has advanced particularly rapidly in recent years and has been deployed widely by technology companies for a range of consumer and business purposes. The report synthesises the state of scientific understanding of general-purpose AI, with a focus on understanding and managing its risks. Amid rapid advancements, research on general-purpose AI is currently in a time of scientific discovery, and – in many cases – is not yet settled science. The report provides a snapshot of the current scientific understanding of general-purpose AI and its risks. This includes identifying areas of scientific consensus and areas where there are different views or gaps in the current scientific understanding. People around the world will only be able to fully enjoy the potential benefits of general- purpose AI safely if its risks are appropriately managed. This report focuses on identifying those risks and evaluating technical methods for assessing and mitigating them, including ways that general-purpose AI itself can be used to mitigate risks.
Other (multiple stages)
Applies across multiple lifecycle stages
Developer
Entity that creates, trains, or modifies the AI system
Manage
Prioritising, responding to, and mitigating AI risks
Primary
7 AI System Safety, Failures & Limitations