FOI release

Response: Participation in the conference on frontier AI safety frameworks

Published 7 October 2024

Reference number FOI2024-00441
Date of request 28 September 2024
Date of response 7 October 2024
Outcome Information disclosed

The Department for Science, Innovation and Technology (the department) received the following request for information which we responded to under the Freedom of Information Act 2000 (FOIA):

Thank you for your prompt response and for publishing some of the information requested. Your answer fails to address some of my request. This is for case [FOI2024-]00412.

Question 1: Under point (1) I asked for the criteria by which attendees were selected, but you have not provided these criteria for the secondary invitees, merely a description. Since there are numerous researchers answering to that description, I am asking for the criteria selecting those who will be selected.

Question 2: Under point (3) I asked for the names of individuals or organisations. Since you are not able to provide names of individuals, please provide the names of the organisations as originally requested.

Question 3: Under point (5) I asked for the arrangements for requesting the opportunity to participate. Your answer merely gives an email address. It seems unlikely that this is the entirety of the arrangement. For example, this email address will presumably be communicated to some target audience of potential participants. Please describe the rest of the arrangements and, as described in point (A) of this email, the criteria you will apply for targeting and selection.

Our response

We can confirm that the Department for Science, Innovation and Technology hold the requested information. We have provided answers to your questions, which are in bold, below:

Question 1

The secondary criteria for invitees to the conference consists of individuals’ prior record of, or current work in, producing high quality work relevant to the topics in the call for submissions.

Across all invitees, we have sought to invite individuals with hands on experience of the technical work or research under discussion such as those doing or who have done the relevant technical research or safety framework development in companies.

Question 2

The signatories to the commitments are researchers from:

  • Ada Lovelace Institute
  • Allen Institute for AI
  • Apollo Research
  • Advanced Research and Invention Agency
  • The Alan Turing Institute
  • Centre for AI Safety
  • Carnegie Endowment for International Peace
  • Centre for Democracy and Technology
  • Centre for Human-compatible AI
  • Cybersecurity and Infrastructure Security Agency
  • Centre for Information Technology Policy
  • City University of London
  • University of York
  • Centre for Long-Term Resilience
  • The Centre for the Study of Existential Risk
  • Centre for Security and Emerging Technology
  • Defence Science and Technology Laboratory
  • Epoch AI
  • EU AI Office
  • Frontier Model Forum
  • Centre for the Governance of AI
  • Haize Labs
  • Hugging Face
  • Krueger AI Safety Lab
  • Model Evaluation and Threat Research
  • Massachusetts Institute of Technology
  • US Nuclear Regulatory Commission
  • Open Philanthropy, Pattern Labs
  • RAND
  • Redwood Research
  • Scale AI
  • Stanford University
  • Tony Blair Institute
  • UC Berkeley
  • Mozilla
  • UK AI Safety Institute
  • US AI Safety Institute
  • Federation of American Scientists
  • Northeastern University
  • AI Now Institute
  • Vector Institute for AI
  • Gray Swan AI
  • Mohamed bin Zayed University of Artificial Intelligence

Question 3

This event is a closed one, primarily held for signatories to the frontier AI safety commitments. The email address is the inbox we use for all public inquiries and correspondence related to the conference.

All expressions of interest in participating and/or making submissions raised with the organisers of the conference (i.e. the UK AI Safety Institute or the Centre for the Governance of AI) are directed to faisc@governance.ai. Such expressions of interest from researchers will be evaluated based on the criteria outlined in response to question A for secondary invitees, taking into account our capacity (within our capacity limit of 90) to extend further invites.

There are no further arrangements in place for individuals to request the opportunity to participate.