International AI Safety Report 2026

Key information

• General-purpose AI systems can perform a wide range of well-scoped tasks with high proficiency. These include conversing fluently in numerous languages; generating code to complete narrow software tasks; creating realistic images and short videos; and solving graduate-level mathematics and science problems.

• However, their capabilities are ‘jagged’: there remain many tasks AI systems do not perform well. For example, AI systems can be derailed by simple errors during multi-step projects; continue to generate text that includes false statements (‘hallucinations’); and cannot yet integrate with robotic components to perform basic physical tasks such as housework. Their performance also tends to decline when prompted in languages other than English, which are less represented in training datasets.

• AI agents are increasingly able to do useful work. For example, AI agents have demonstrated the ability to complete a variety of software engineering tasks with limited human oversight. However, they cannot yet complete the range of complex tasks and long-term planning required to fully automate many jobs.

• Since the publication of the last Report (January 2025), advances in ‘reasoning systems’ have driven performance improvements on more complex tasks. Reasoning systems are able to break problems into smaller steps and compare alternative answers. This has especially improved their performance on tasks related to mathematics, coding, and scientific research.

• A central challenge is an emerging ‘evaluation gap’: existing evaluation methods do not reliably reflect how systems perform in real-world settings. Many common capability evaluations are outdated, affected by data contamination (when AI models are trained on the same questions used in evaluations), or focus on a narrow set of tasks. As a result, they provide limited insight into real-world AI performance.

General-purpose AI systems exhibit many remarkable capabilities. Leading systems now perform at gold-medal level in mathematics competitions and assist scientific researchers with generating hypotheses and troubleshooting laboratory work. They match, and in some cases exceed, expert performance on a wide range of benchmarks and task-specific evaluations. Yet the performance profile these systems display is also ‘jagged’: their capabilities vary widely among different tasks and contexts. They still sometimes generate false information (‘hallucinations’) and produce inconsistent outputs even when given identical or similar inputs. An ‘evaluation gap’ exists: AI systems often perform impressively in controlled settings such as pre-deployment evaluations, but more poorly in real-world conditions. This variability makes it difficult to assess general-purpose AI capabilities with a single metric. This section outlines both the capabilities of AI systems and their shortcomings (Table 1.4).

What can current general-purpose AI systems do?

General-purpose AI systems now perform at or above the level of human experts on standardised evaluations, covering a growing range of well-defined professional and scientific subjects (Figure 1.4). For example, leading models score over 90% on undergraduate-level examinations in subjects from chemistry to law (MMLU,) and achieve over 80% on graduate-level science tests (GPQA,). In July 2025, models from Google DeepMind and OpenAI reached gold medal-level scores at the International Mathematical Olympiad, solving five out of six problems under competition-like conditions. Beyond text-based reasoning, these systems display powerful multimodal capabilities: they can create photorealistic images, short high-definition videos, 3D scenes, and musical pieces from simple text prompts, and they are beginning to process complex sensor data to guide physical robots.

Advanced capabilities are increasing productivity in medicine, education, software development, and other sectors

Advanced AI capabilities now power practical tools that match or exceed human performance on specific tasks, increasing productivity in multiple sectors.

• In medicine, AI systems can analyse clinical scenarios and conduct diagnostic conversations to generate lists of potential diagnoses. In specific simulated settings, their accuracy can exceed that of human physicians, though they lack the reliability and consistency required for real-world clinical deployment.

• In education, AI systems are being rapidly adopted in areas from curriculum design to student assessment, transforming the education process, while widespread use by students is posing significant challenges to the integrity and validity of existing academic assessments.

• In software development, AI coding assistants are now widely adopted, with some studies suggesting that developers using AI assistants complete certain tasks 20–30% faster on average than those without.

• Large-scale studies in other sectors such as customer service, consulting, and professional writing find measurable productivity gains from AI-assisted work, though these effects vary across tasks and worker groups. (For a more detailed discussion of the labour market implications of general-purpose AI, see §2.3.1. Labour market impacts.)

General-purpose AI systems assist scientific research

General-purpose AI systems are now used by researchers to support relatively complex tasks across disciplines. Researchers have demonstrated that AI systems can, under high-level human guidance, design novel proteins for medical use, which are later validated in a physical laboratory. Other systems have discovered new algorithms that are more efficient than long-standing human-designed methods. Notably, such advances often rely less on the raw power of the latest models and more on appropriate system integration. General-purpose AI is also increasingly used to accelerate AI research itself, a trend with significant implications discussed further in §1.3. Capabilities by 2030. In the social sciences, researchers are using AI to accelerate data analysis through automated annotation and to explore social dynamics by simulating individual and collective behaviour with AI agents. Moving from analysis to direct application, researchers are beginning to use general-purpose AI systems to design and study scalable, novel social interventions. For example, recent work has explored using AI-mediated conversations to find common ground in democratic debates or to reduce belief in conspiracy theories through dialogue.

What are the current limitations of general-purpose AI systems?

Despite advances in capabilities, the performance of general-purpose AI systems remains jagged across tasks and contexts. This section highlights some prominent limitations, though the full range of challenges is broader.

Reliability challenges persist in current AI systems

Despite recent improvements, general-purpose AI systems can be unreliable and prone to basic errors of fact and logic. Even systems that excel at complex tasks may generate non-existent citations, biographies, or facts – a phenomenon known as ‘hallucination’. Their performance can also be inconsistent; for example, accuracy on maths problems can decrease significantly when irrelevant information is inserted into the problem description. This brittleness extends to multimodal capabilities, where models often have low performance on spatial reasoning tasks, such as basic counting of objects in a scene. While expert human oversight can mitigate some of these risks, there is a corresponding danger of over-reliance, where users trust incorrect outputs because they are presented fluently and confidently (see §2.3.2. Risks to human autonomy). This unreliability makes it difficult to safely adopt such systems in high-stakes settings such as medicine and finance, where errors can have grave consequences, and human verification of system outputs remains necessary.

Systems struggle with long-term planning and unexpected obstacles

General-purpose AI systems also struggle with tasks that require long-term planning, maintaining a coherent strategy over many steps, and adapting to unexpected obstacles. As tasks grow longer, AI agents often lose track of their progress and cannot reliably deal with unexpected inputs. For example, even a simple website pop-up ad can derail an entire task. Large-scale evaluations confirm this pattern: in software development, the most capable systems achieve only 50% success on tasks lasting just over two hours, and reaching 80% success requires limiting them to much simpler 25-minute tasks. For now, reliable automation of long or complex tasks remains infeasible.

Interacting with the physical world remains challenging

Progress on digital tasks has also proved difficult to translate into robotics, where the complexity of the physical world introduces new challenges. Recent advances are centred on Vision-Language-Action (VLA) models – foundation models designed to enable robots to follow natural language instruction, interpret multimodal sensory data, and generate motor commands. State-of-the-art systems such as π0.5 and Gemini Robotics can now interpret simple verbal commands such as ‘clean the kitchen’ and execute a sequence of physical steps in an unfamiliar, controlled environment. However, current VLA models still do not perform well with unusual object shapes and unexpected events. Ensuring that such systems can operate safely and reliably to minimise the risk of physical harm or property damage, and perform well in diverse environments remains an active area of research.

Performance is uneven across languages and cultures

The capabilities of general-purpose AI models and systems also vary across languages and cultures. Performance is highest on tasks in English, reflecting the fact that most training data comes from Western sources. For example, one evaluation of AI models across 83 languages found substantially lower performance on languages that use non-Latin scripts and on languages with limited digital resources. This disparity extends to cultural knowledge; in one study, AI models correctly answered 79% of questions about everyday US culture but only 12% of questions about Ethiopian culture. Another study finds that current models ‘reason’ more effectively in high-resource languages, which may widen the performance gap between languages. Beyond language and culture, similar patterns appear along geographic and socioeconomic lines. Models underrepresent locations with disadvantaged demographics in recommendations – for example, if asked for a restaurant recommendation, they might fail to suggest restaurants in poorer areas – and their performance on factual recall degrades for lower-income countries. This inequality is compounded by evaluation benchmarks that are themselves heavily skewed toward English, creating an ecosystem where low-resource languages remain systematically understudied and underoptimised.

Updates

Since the publication of the last Report (January 2025), ‘reasoning’ systems have become mainstream (see §1.1. What is general-purpose AI? for details of their development). These systems demonstrate substantially improved performance on hard mathematics, coding, and scientific tasks by generating and comparing multiple answers within their own ‘chain of thought’ before producing a final answer (Figure 1.5). Because these models’ performance depends in part on inference-time compute, their effective capabilities can change after initial development – improving as more computational resources are allocated.

In parallel, AI companies have focused more on developing AI agents, especially in areas such as software engineering and computer use. While reliability remains a bottleneck, the complexity of tasks these agents can automate has been increasing rapidly. Finally, enabling models to form long-term memories and learn continuously from user interaction is emerging as a key area of development.

Evidence gaps

Jagged capabilities and the evaluation gap make general-purpose AI capabilities difficult to reliably measure and predict. Performance also depends heavily on the specific test examples and prompt used, making it difficult to prove with high-confidence that an AI system cannot perform certain – potentially dangerous – tasks. There is no single, comprehensive, and continuously updated synthesis of AI capabilities, leading to a fragmented and often outdated understanding of the field. Existing reviews, including this Report, provide valuable summaries but are static snapshots in a rapidly moving field. With no widely accepted taxonomy for capabilities, policymakers must navigate a patchwork of benchmarks and sources to form a complete picture.

Benchmarks often fail to predict real-world performance

Benchmark integrity is a growing concern. Many capability evaluations rely on standardised benchmarks. However, many models may have been trained using data from these same benchmarks – a problem called ‘data contamination’, which most developers do not currently track or disclose. This can lead to inflated performance scores that do not reflect a model’s true ability, but rather its capacity to memorise answers. A further limitation of current evaluation practices is that they rely on automated testing in controlled lab environments. However, this often overestimates AI systems’ practical utility in dynamic, real-world settings. For example, one study found that, while an AI agent could produce functional code, the code still required significant human effort to fix issues with documentation, formatting, and quality before it was usable in a real project. To address these limitations, a dedicated ‘evaluation science’ is emerging, advocating for rigorous methodologies that ensure external validity and better predict real-world performance. For instance, recent benchmarks have begun to measure AI system performance on economically valuable tasks and real-world remote labour.

The evidence for how AI augments human capabilities is inconclusive

Measuring AI’s practical benefits consistently is challenging because success depends on both the specific task and the user’s skill at leveraging AI for it, meaning lab results often fail to predict real-world value. For example, one study shows that a model’s standalone accuracy is not a reliable predictor of human-AI team performance. Many studies confirm positive uplift from using AI. However, one recent study found that, although software developers believed that AI was making them more productive, it actually slowed down experienced programmers by 19% on complex coding tasks.