On Friday, 17 April 2024, new research published by the Department for Science, Innovation and Technology (DSIT), the AI Safety Institute (AISI), and supported by over 30 nations, including EU and UN representatives, underscores the potential impact of artificial intelligence (AI) if collaboration on AI safety is not deepened.
The International Scientific Report on the Safety of Advanced AI, which was launched at the first Global AI Safety Summit at Bletchley Park, aims to provide policymakers worldwide with a comprehensive source of information on AI safety. While advanced AI offers various benefits, such as advancements in healthcare and climate change solutions, it also poses potential risks, including disinformation campaigns and economic imbalances. The report also highlights the lack of consensus among AI experts on various topics, emphasising the need to prioritise understanding and decision-making for the future. The final report, expected to be published for the AI Action Summit to be hosted in France, will incorporate industry, civil society, and the AI community feedback to reflect the latest research and ensure a comprehensive view of advanced AI risks.
Among the areas researched in the report, Infosecurity Magazine highlights that four out of five of the most widely used generative AI chatbots are highly susceptible to basic jailbreak attempts. The AISI conducted tests to evaluate the cyber risks associated with these models, including their vulnerability to jailbreaks, potential use in facilitating cyber-attacks, and their autonomous capabilities. The results indicated that all of the chatbots were highly vulnerable to basic jailbreak techniques, with the models producing harmful responses in 90% to 100% of cases when exposed to repeated attack patterns.
In related news, DSIT has announced that the AISI is looking to open offices in San Francisco. The expansion will enable the UK to access the world's leading talent and AI labs.
Then on Wednesday, 22 May 2024, DSIT announced £8.5 million in AI grants will be available to researchers working on new methods to increase safe and trustworthy deployment of AI Funded research.
What is this page?
You are reading a summary article on the Privacy Newsfeed, a free resource for DPOs and other professionals with privacy or data protection responsibilities helping them stay informed of industry news all in one place. The information here is a brief snippet relating to a single piece of original content or several articles about a common topic or thread. The main contributor is listed in the top left-hand corner, just beneath the article title.
The Privacy Newsfeed monitors over 300 global publications, of which more than 4,350 summary articles have been posted to the online archive dating back to the beginning of 2020. A weekly roundup is available by email every Friday.
