SaferAI

SaferAI is a French nonprofit (association loi 1901) founded in 2023 by Simeon Campos, who previously co-founded EffiSciences, an organization providing training in responsible AI. Headquartered in Paris, SaferAI works across research and policy to advance responsible AI development by modeling complex risks, improving company practices, and contributing to standards and policies. The organization's work spans three main pillars. First, SaferAI conducts fundamental research developing quantitative risk models that translate AI capabilities into measurable real-world harm assessments. The team has completed nine cyber risk models examining how AI could amplify cyberattack effectiveness, and is expanding into loss of control risks, CBRN threats, and manipulation scenarios. This work draws on established risk management practices from industries like aviation and finance. Second, SaferAI independently evaluates and rates leading AI companies' risk management maturity through its public ratings platform (ratings.safer-ai.org). Companies are assessed across four dimensions: risk identification, severity assessment, safeguard implementation, and system governance. The first ratings, released in 2024, found that even industry leaders like Anthropic scored only 2.2 out of 5 ("moderate"), while companies like xAI scored 0. These ratings received significant media coverage in TIME and Euractiv and attracted attention from institutional investors including Norway's sovereign wealth fund. Yoshua Bengio endorsed the ratings as crucial for independent assessment. Third, SaferAI contributes extensively to international AI governance standards. The team leads AI risk management standards development in CEN-CENELEC JTC 21, edits ISO generative AI red teaming standards, helped shape the EU AI Act Code of Practice for general-purpose AI models, and contributed to the OECD Hiroshima AI Process reporting framework. SaferAI is also part of the US AI Safety Consortium (NIST AISIC) and advises AI Safety Institutes globally including the UK AISI. The team includes about 11 staff members with diverse backgrounds spanning ML research, risk management, cybersecurity, policy, and standards development. The research team is led by Malcolm Murray, formerly Chief of Research at Gartner, and includes PhD-level researchers in computer science, particle physics, and political economy. Senior advisors include Cornelia Kutterer (former Senior Director of EU Government Affairs at Microsoft), Duncan Cass-Beggs (former OECD head of strategic foresight), and Fabien Roger (Technical Staff at Anthropic). As of late 2025, founder Simeon Campos began transitioning out of the Executive Director role to start a new venture, with Henry Papadatos taking over organizational leadership.