AI Safety Summit - Enhancing Frontier AI Safety (Amazon)

• Authority: Private-sector companies
• Collections: Corporate policies and commitments
• Most recent activity: Enacted
• Most recent activity date: 2023-10-01

Risk subdomain coverage

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

• Unfair discrimination and misrepresentation
• 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
• 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.

• Unequal performance across groups
• False or misleading information
• Overreliance and unsafe use
• Competitive dynamics
• Governance failure
• AI pursuing its own goals

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), lack of capability/robustness (7.3), lack of transparency (7.4), and governance failure (6.5). Coverage is concentrated in security, misuse prevention, AI safety evaluation, and responsible AI practices.

Sector coverage

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

• Information
• Scientific Research & Development Services

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

• Trade, Transportation & Utilities
• Educational Services
• Accommodation, Food & Other Services

All other sectors had no coverage in this document.

Sectors Summary

As an internal corporate policy document, this primarily governs Amazon's operations in the Information sector (cloud services, data processing) and Scientific Research and Development Services sector (AI/ML research and development). The document also references AI applications across multiple sectors that AWS customers operate in, but does not directly govern those sectors.

Legislative Status

Other

Legislative Status Summary

This is an internal corporate policy document describing Amazon's AI governance framework, responsible AI practices, and voluntary commitments. It lacks binding legal obligations, enforcement mechanisms, or penalties characteristic of hard law.

Actors

Proposer

• Entity Types: AI_Developer, AI_Infrastructure_Provider
• Entities: Amazon; Amazon Web Services (AWS)
• Summary: Amazon/AWS authored this document describing their AI governance framework, responsible AI practices, and safety commitments. The document represents Amazon's internal policies and voluntary commitments as both an AI developer (Titan models) and infrastructure provider (AWS services).

Target

• Entity Types: AI_Developer, AI_Deployer, AI_User
• Entities: AWS customers; Amazon internal teams; third-party model providers
• Summary: The document applies to multiple audiences: internally to Amazon's AI development teams and practices, to AWS customers who use Amazon Bedrock and other AI services, and to third-party model providers whose models are available on AWS platforms.

Enforcer

• Entity Types: AI_Developer, AI_Infrastructure_Provider
• Entities: AWS Trust & Safety team; Amazon Security
• Summary: AWS enforces its responsible AI policies through internal teams including the Trust & Safety team and Amazon Security, with authority to suspend service access and investigate policy violations.

Monitor

• Entity Types: AI_Developer, AI_Infrastructure_Provider
• Entities: AWS Security Operations Center; AWS Trust & Safety team; AWS Penetration Testing Program
• Summary: AWS monitors AI systems through multiple mechanisms including Security Operations Centers, automated abuse detection, continuous testing, and the Trust & Safety team that tracks policy violations and emerging risks.

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 covers all stages of the AI lifecycle comprehensively, with particularly strong emphasis on Build and Use Model, Verify and Validate, Deploy, and Operate and Monitor stages. It describes Amazon's end-to-end approach from data collection through post-deployment monitoring.

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, AI systems, and frontier AI. It extensively discusses foundation models and generative AI, with specific focus on Amazon's Titan Foundation Models. The document does not explicitly define compute thresholds or distinguish between general purpose and task-specific AI, though it focuses on general-purpose foundation models. Open-weight models are briefly mentioned in context of risk assessment.