SaferAI

SaferAI's independent evaluation of GLM-5.2, the first in Europe, tests Zhipu AI's open-weight flagship across the four systemic risk areas in the EU Code of Practice and finds frontier-level capability on cyber and biology benchmarks without the safeguards frontier developers ap

Although general-purpose AI systems offer transformational opportunities in science and industry, they simultaneously raise critical concerns about safety, misuse, and potential loss of control. Despite these risks, methods for assessing and managing them remain underdeveloped. E

Rapidly advancing artificial intelligence (AI) systems introduce novel, uncertain, and potentially catastrophic risks. Managing these risks requires a mature risk management infrastructure whose cornerstone is rigorous risk modeling. We conceptualize AI risk modeling as the tight

This memo conducts a structured comparison between emerging Frontier Safety Frameworks and established international risk management standards, and argues that blending the operational specificity of FSFs with the rigor and maturity of global standards offers a promising pathway

In this research memo, the Oxford Martin AIGI brings together diverse stakeholders to sketch out the contours of a “gold-standard” framework for risk tiering—balancing quantitative and qualitative assessments, lifecycle classification, and the integration of benefit-risk reasonin

This report contains an analysis of the commonalities and differences between the G7 Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems and the EU AI Act text on general-purpose AI models. There is substantial commonality between the

The literature and multiple experts point to many potential risks from large language models (LLMs), but there are still very few direct measurements of the actual harms posed. AI risk assessment has so far focused on measuring the models' capabilities, but the capabilities of mo

The recent development of powerful AI systems has highlighted the need for robust risk management frameworks in the AI industry. Although companies have begun to implement safety frameworks, current approaches often lack the systematic rigor found in other high-risk industries. T

SaferAI argues that evaluation results only improve governance if there is a pipeline to interpret and act on them, and describes its framing around multiple inputs to risk modeling.

SaferAI published an independent evaluation of Zhipu AI’s open-weight flagship GLM-5.2 across four systemic risk areas aligned with the EU General-Purpose AI Code of Practice.

SaferAI published an external review of DeepMind’s public “safety case” for scheming inability, describing gaps and recommendations for how external review of frontier AI safety cases should work.

SaferAI published a loss-of-control risk analysis framed via systems-thinking approaches.

SaferAI published a repository-style synthesis identifying stronger current practices for frontier AI risk management across risk identification, analysis/evaluation, treatment, and governance.

SaferAI argued for technologies the EU could fund to make AI reliability research a reality, describing an economic/institutional rationale and references to a coalition for AI assurance.