Some top AI labs have ‘very weak’ risk management, study says

According to a new study by the French nonprofit SaferAI, many leading AI labs, including Elon Musk’s xAI, are lacking in key safety measures. Siméon Campos, founder of SaferAI, said: “AI is extremely fast-moving technology, but AI risk management isn’t moving at the same pace.” The study’s ratings, which assess risk management practices, show that xAI received a score of 0/5, the lowest amongthe six companies evaluated. Meta and Mistral AI were also rated as having “very weak” risk management, while OpenAI and Google Deepmind received “weak” ratings. Campos hopes that the ratings will encourage companies to enhance their internal processes to reduce bias and misinformation. The grading framework aligns with international AI standards, including the EU AI Act. Yoshua Bengio, a prominent AI figure, endorsed the ratings, emphasizing the need for accountability in AI development.