2023 UK AI Safety Summit Response (Meta)

• Authority: Private-sector companies
• Collections: Corporate policies and commitments
• Most recent activity: Proposed
• Most recent activity date: 2023-10-20

Risk subdomain coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the risk subdomain are explicitly described.

• Exposure to toxic content
• Compromise of privacy
• AI system security vulnerabilities
• Disinformation, surveillance, and influence at scale
• Cyberattacks, weapon development or use
• Fraud, scams, and targeted manipulation
• AI pursuing its own goals
• AI possessing dangerous capabilities
• Lack of capability or robustness
• Lack of transparency or interpretability

Minimal coverage — Brief mention but no specific mitigations, recommendations or governance measures relating to the risk subdomain are described.

• Unfair discrimination and misrepresentation
• False or misleading information
• Overreliance and unsafe use
• Competitive dynamics

All other risk subdomains had no coverage in this document.

Risk Coverage Summary

The document has good coverage of approximately 10-12 subdomains, with strong focus on AI system security (2.2), malicious actors and misuse (4.1, 4.2, 4.3), privacy compromise (2.1), AI safety failures (7.1, 7.2, 7.3), lack of transparency (7.4), and competitive dynamics (6.4). Coverage is concentrated in security, misuse prevention, AI safety, and responsible development domains.

Sector coverage

Good coverage — Comprehensive governance measures or mitigations with clear procedures relating to the sector are explicitly described.

• Information
• Scientific Research & Development Services

All other sectors had no coverage in this document.

Sectors Summary

This is an internal corporate policy document from Meta, an AI developer company. The primary sector governed is Information (where Meta operates as a social media, technology, and AI development company). The document does not regulate external sectors but describes Meta's own responsible AI development practices.

Legislative Status

Other

Legislative Status Summary

This is an internal corporate policy document representing Meta's voluntary commitments and responsible AI practices. It describes Meta's approach to AI safety and references their adherence to the White House Commitments, but contains no binding legal obligations, enforcement mechanisms, or penalties.

Actors

Proposer

• Entity Types: AI_Developer
• Entities: Meta; Fundamental AI Research (FAIR) team; Responsible AI Team
• Summary: Meta is the author and proposer of this governance document, describing their own responsible AI development practices and commitments. The document is authored by Meta as an AI developer company.

Target

• Entity Types: AI_Developer
• Entities: Meta; developers using Llama 2; Meta's internal teams
• Summary: The document primarily targets Meta's own operations and development practices. It also provides guidance for external developers who use Meta's Llama models, making them secondary targets.

Enforcer

• Entity Types: AI_Developer
• Entities: Meta's internal review teams; Youth Advisory Council; Safety Advisory Council
• Summary: Meta enforces its own policies through internal review processes, advisory councils, and monitoring mechanisms. There is no external enforcement body mentioned.

Monitor

• Entity Types: AI_Developer, AI_User, Affected_Stakeholder
• Entities: Meta; red teamers; community users; external researchers; UC Berkeley; NIST
• Summary: Monitoring is conducted by Meta internally, external red teamers, community users providing feedback, and academic researchers. The document describes multiple monitoring mechanisms including bug bounties and feedback systems.

AI Lifecycle Stages

Lifecycle stages covered by the documents, as defined in the OECD Framework for the Classification of AI Systems

• Plan and Design
• Collect and Process Data
• Build and Use Model
• Verify and Validate
• Deploy
• Operate and Monitor

AI Lifecycle Summary

The document comprehensively covers multiple AI lifecycle stages with particular emphasis on Build and Use Model, Verify and Validate, Deploy, and Operate and Monitor. It describes extensive model development, testing, deployment practices, and post-deployment monitoring mechanisms.

Technical Scope

Terms relating to technical scope that are mentioned or implicitly referenced in the document

• AI Models
• Foundation Models
• Frontier AI
• AI Systems
• Generative AI
• Open Weight / Open Source

Technical Scope Summary

The document explicitly covers AI models and AI systems, with specific focus on frontier AI and foundation models (Llama 2). It addresses generative AI extensively and mentions open-weight models. There is no explicit mention of compute thresholds, general purpose AI, task-specific AI, or predictive AI.