We re-re launched the ILINA blog with Marie’s excellent piece. I learnt a lot reading parts of it as she worked on it, and you will too. Also, great concrete suggestions ilinaprogram.org/blog/leveragin…

re: incident monitoring + reporting in Africa - best so far I’ve read! Highly recommend from a solid AI safety/ gov researcher!✨🙌🏾

New personal blog post on something I’ve been thinking about on and off for a while now: Africans need to Reclaim Pan Africanism in the Age of AI. Read, share 🙏🏾 cecilyongo.medium.com/africa-needs-t…

Our Junior Research Scholar Gathoni Ireri has been working on the risk of AI manipulation. In a new piece in Tech Policy Press, she takes an in-depth look at what long-term memory in ChatGPT, Gemini & others means for autonomy. Read her work here: techpolicy.press/what-we-risk-w…

Gathoni Ireri is at the forefront of our work on AI manipulation, and you can now read her first piece on the subject. Very proud of her ✨ Lots more excellent and field-leading work on the way!

A lot of my work at ILINA this year is focused on manipulation risks from AI. There are a bunch of threads you can pull, and it is, for me, an endlessly fascinating question. In this piece I pulled on one asking: what does it mean now that LLMs have cross-conversational memory?

Noni’s new piece marks the 10th publication by one of our team members. Small milestone but lovely nonetheless. Each of the pieces there is a result of numerous hours of research, discussion, writing, rewriting, disappointment and delirium. Lots more on the way!

We were delighted to be joined for our Work-in-Progress meeting today by Elizabeth Cooper, Director of Berkeley Existential Risk Initiative, an org that graciously supports our work. We discussed new research by Cecil Abungu and Elvis Mogesa and then discussed upcoming work.

We're excited to announce the winners of our AI Agents & the Law course project proposal competition!

Pleased that my proposal won 1st place in Vista Institute's AI Agents & the Law proposal competition.🏅✨ Hopefully this project informs how courts weigh evals as evidence. Read an analysis of it by Cristina Ibáñez and Dr. Aishwarya Saxena here: vistainstituteai.substack.com/p/how-courts-w…

Congratulations to our Junior Research Scholar Mark Lenny for winning first place in Vista Institute for AI Policy's course project competition 👏 Mark's win is a testimony to his hard work and the strength of the ongoing talent development work at ILINA. We are very proud!

😁😁 Huge move, love it!!💯💯

Our Junior Research Scholar Raqda has been researching liability regimes for frontier AI. You can now read her new blog post "What the Megan Garcia Case Tells Us About AI Liability in the U.S." in the RAILS e.V. blog: blog.ai-laws.org/what-the-megan…

Most of my work at ILINA this year has been centered on frontier AI liability. The Megan Garcia case is the first US case to sketch out the contours of AI liability in the US. In this piece I break down some of the key findings in the case and their immediate implications.

Today the ILINA team had an extensive collaboration meeting with the Action Lab Africa team. We're looking forward to working together on topics like AI safety evaluations and AI safety policy engagement in Africa.

Today we hosted a Work-in-Progress session in which our Pre-Fellow Yusra Abdullahi presented her upcoming work, which focuses on how the use of AI at law clinics requires guardrails.

Congrats to Yusra A. for having enough courage to present her research to our team today👏🏾 Looking forward to seeing her final paper!

Everyone here was such huge help. They helped me rethink my work and reorganise my ideas. Super grateful for ILINA!!! Also as much as it was nerve wracking, I feel like the feedback was for the better. Highly recommend‼️😤

Our intensive 10 week course on machine learning and deep learning ended today with a Football Friday tribute to our excellent course facilitators, Carringtone Kinyanjui and Roger Odipo ⚽️😁✨

This was the most demanding class I've taught this year and to a very intelligent, patient, kind ( and naughty! ) class More please 😊