Cover Image for How do we solve alignment?

Presented by

Hub for all Meridian events: AI safety research talks, biosecurity programming, policy discussions, fellowship activities, and community gatherings. Based in Cambridge, focused on frontier risk.

Hosted By

21 Went

AI

How do we solve alignment?

Name: How do we solve alignment?
Start: 2026-03-16T17:30:00.000+00:00
End: 2026-03-16T18:30:00.000+00:00
Location: Meridian Cambridge

Meridian

Meridian Cambridge

Cambridge, United Kingdom

Sold Out

This event is sold out and no longer taking registrations.

About Event

How Do We Solve Alignment?

A technical AI safety speaker series hosted by CAISH at Meridian, Cambridge.

This week: Leon Staufer: The 2025 AI Agent Index

Leon will present the 2025 AI Agent Index, which documented 1,350 fields across 30 state-of-the-art AI agents. Some highlights: 25 out of 30 agents disclose no internal safety results and only 4 have agent-specific safety evals. Nearly the entire ecosystem backends to just three model families (GPT, Claude, Gemini). Only 6 out of 30 agents respect robots.txt, and most don't identify themselves as bots online. There are also notable US/China gaps in transparency and safety reporting.

The core argument: without better disclosure standards and shared protocols for how agents behave on the web, governing agents is going to be very difficult.

Full index, data, and paper at https://aiagentindex.mit.edu

Attendance requires approval. When you register you'll be asked about your background in AI safety. We will be providing food!

Location