🇪🇺 Also, the EU AI Act is now *officially* final, and enters into force today! Check out our new implementation timeline for a list of implementation milestones through 2031.

Our New $4 Million Grants Program

With ungoverned AI development currently on track to concentrate power among a small number of groups, individuals, organizations, and corporations, we’ve launched a new grants program of up to $4 million to support projects working to mitigate this, and move us towards a better world of meaningful human agency.

Power concentration here could refer to the “ownership of a decisive proportion of the world’s financial, labour, or material resources, or at least the ability to exploit them. It could be control of public attention, media narratives, or the algorithms that decide what information we receive.”

Applications are being accepted on a rolling basis. The first round of review for projects began this week; the second will commence 15 September. Find out more, and apply, here.

Pindex x FLI Collaboration

With our partners at Pindex, we helped to produce a popular new video with iconic actor Stephen Fry outlining the startling risks from the continued development of - and investment into - evermore powerful AI systems.

Watch and share it below:

Problem-Solving AI Grant Recipients

We’re Hiring!

We’re hiring for four roles across FLI:

1. U.S. Communications Manager (Permanent, Full-Time)

2. Operations Associate (Permanent, Full-Time)

3. Video Producer (Contract, Part-Time)

4. Content Writer and Researcher (Contract, Part-Time)

All roles support remote work. Apply by 4 August at the links above!

Updates from FLI

• FLI’s U.S. Policy Specialist Hamza Chaudhry wrote an op-ed in TIME, about the under-discussed risks of the “near-complete absence of [AI safety and capabilities] testing in non-English languages.”

• Hamza was quoted in another TIME article on Meta’s recent release of its new open-source LLM:

• Dr. Emilia Javorsky, Director of FLI’s Futures Program, had an essay published in Noema explaining why we need an “FDA for AI”.

• Executive Director Anthony Aguirre released a statement calling for stronger whistleblower protections and AI regulation in the U.S., in light of recent allegations from OpenAI employees.

• Anthony was also cited in this New York Times article on AI “slop”.

• FLI’s AI Safety Summit Lead Ima Bello hosted AI pioneer Stuart Russell for the launch of the Paris AI Safety Breakfast series:

• FLI’s EU Research Lead Risto Uuk spoke to Positive.News about increased cyber risks from AI.

• Our Superintelligence Imagined contest, offering five $10,000 prizes for the best creative materials on risks from superintelligence, was covered by Graphic Competitions, InfoDesigners, FAD Magazine, and ArtInfo.

• Semafor covered our new AI-driven power concentration grants program in their tech newsletter.

• The Foresight Institute published a blog post outlining the three winning AI governance institution proposals from the Existential Hope Hackathon we co-hosted in February.

• We were honoured to have Mary Robinson, former President of Ireland and Chair of The Elders (with whom we’re calling for long-view leadership addressing existential threats) on the FLI podcast. She and host Gus Docker discussed how to overcome barriers to international cooperation, her advice to the next generation of world leaders, long-view leadership, and more.

• FLI’s Dr. Emilia Javorsky also joined the podcast for an episode on how AI concentrates power, how we might mitigate that, and what utopia(s) could look like.

What We’re Reading

• Working through anti-AI safety arguments: AI expert and Turing Award winner Yoshua Bengio published an extensive blog post, reasoning through 12 of the most popular arguments against taking AI safety seriously. Read his full post, and check out our summary of it below:

• “Facebook’s anti-regulatory attack dog”: Transformer News’ Shakeel Hashim highlights Meta’s expensive advertising efforts - via a “dark money group” - against AI regulation, seemingly appealing to fears about U.S.-China competition.

• A race the public doesn’t want: Speaking of U.S.-China competition, new polling finds that 75% of Americans don’t buy into the argument that U.S. AI companies should race ahead building powerful systems, without regulations, in order to compete with China - a narrative that many of the companies themselves are evidently trying to push.