A new interim report on advanced AI safety, compiled by experts from 30 countries, including the UK, US, and China, highlights potential risks and benefits of general-purpose AI. Led by Professor Yoshua Bengio, the report identifies key risk categories and will shape future discussions on AI safety.
A new report on advanced AI safety by experts from 30 countries, including the UK, US, and China, was released recently. The International Scientific Report on Advanced AI Safety identifies both the potential benefits and risks associated with general-purpose AI. This interim report, the first iteration, was compiled by independent experts without significant input from AI companies.
The report, led by AI expert Professor Yoshua Bengio, was introduced at the UK-led AI Safety Summit at Bletchley Park in November 2023. It highlights three main risk categories: malicious use, such as scams and deepfakes; malfunctions, including systemic bias and loss of control; and systemic risks like job market disruptions and privacy issues.
The future of general-purpose AI is described as “uncertain,” with both beneficial and harmful possibilities. The findings will be further discussed at the upcoming AI Seoul Summit in South Korea.
This report will inform ongoing global discussions about AI safety, with a final version expected before the AI Action Summit in France.
