A provisional AI safety report has highlighted the lack of consensus among AI experts on various topics, including the current state of AI capabilities and their potential future developments. The report, part of the Bletchley Declaration, examines differing opinions on extreme risks such as widespread unemployment and AI-enabled terrorism. The experts involved agreed that society and policymakers need a better understanding of AI’s impact. The report’s chair, Yoshua Bengio, emphasized the need for responsible use, development, and regulation of AI. The report will inform discussions at the AI Seoul Summit and will be updated with input from industry and civil society. The final report is expected to be published in time for the AI Action Summit in France.