For example, research shows that:
· When given 10 attempts, attackers can use malicious prompts to bypass leading systems' safeguards about half the time.
· Inserting as few as 250 malicious documents into a model's training data can introduce vulnerabilities.
(6/6)

Yet significant challenges remain and the real-world effectiveness of many safeguards is uncertain. ⬇️
(5/6)

· A growing number of companies adopting Frontier AI Safety Frameworks, describing safety and security measures they will take as their AI models become more capable,
· Technical safeguards are beginning to inform transparency measures in governance frameworks.
(4/6)

Over the past year, we’ve seen meaningful progress:
· Improvements in adversarial training methods to make models more resistant to potentially harmful requests,
· Better tools for tracking AI-generated content.
(3/6)

This key update, paired with the first one published in October, aims to provide policymakers and researchers with timely information about important trends in frontier AI development.
Read the full Update here: internationalaisafetyreport.org/publication/...
(2/6)

Learn more about the 5 main concrete actions they should pursue in our latest Blueprint for Multinational Advanced AI Development: aigi.ox.ac.uk/publications...

It was equally a pleasure to receive this distinction alongside an exceptional group of colleagues, whose contributions have had a profound impact on the field of AI as we know it today. www.youtube.com/watch?v=0zXS...

Addressing these risks doesn't mean stopping progress.
Innovation & regulation must go hand in hand, notably by building technical safeguards to make AI systems more trustworthy for individuals and businesses. That's at the heart of our work at @law-zero.bsky.social.

Thank you to the University of Copenhagen, the European Commission @ec.europa.eu, and my co-panelists — Lene Oddershede, Max Welling, Peter Sarlin & @fabiantheis.bsky.social — for a day of excellent discussions.

I want to congratulate EVP @hennavirkkunen.bsky.social and EU Commissioner Ekaterina Zaharieva on the launch of RAISE, a critical initiative to support AI for Science and Science for AI.

We covered current evidence around AI's potential for misuse, governance & international cooperation mechanisms, and strategies to ensure AI security.