An AI safety company that develops and operates large language models, including the Claude family of AI assistants used by consumers and businesses for text generation, analysis, and conversation. The company's policies govern how user data is collected and processed during AI interactions, as well as what types of content and use cases are prohibited on their platforms. These policies are significant for users because they determine privacy protections for sensitive conversations with AI systems and establish boundaries for acceptable AI-assisted activities.
Higher = more protective of you, the user/customer — a deliberately user-protective lens, not a universal “good vs bad” score. Computed algorithmically from Anthropic’s cited governance stances; no dimension we haven’t confidently read is ever counted against the score. Free, reproducible, and never purchasable.
Anthropic's authority to pause development is not constrained to RSP-mandated triggers, meaning voluntary action is available under a broader set of self-assessed circumstances.
Because no notice is required, a user's access to the Services can be altered or eliminated at any moment without warning, with no procedural protection.
Facial geometry templates are among the most sensitive categories of personal data and are regulated as biometric data in several jurisdictions; their collection by Anthropic carries significant priv…
A requirement to quantify risk across all deployed models — rather than selected ones — establishes a comprehensive scope for Anthropic's internal risk accounting obligations.
The requirement directly restricts who can access model weights and under what conditions, establishing a structural safeguard against unauthorized or unilateral access to core AI assets.
This document lists the outside companies Anthropic sends user data to, what each company does with it, and which Anthropic products are involved. Every Anthropic product routes data through at …
This is Anthropic's system card for Claude Sonnet 4.5, a technical disclosure document describing how the model is built, evaluated, and constrained. The document states that certain behaviors are hardcoded …
This is Anthropic's system card for Claude Opus 4.8, a large language model released in May 2026, documenting pre-deployment safety evaluations across chemical and biological weapons risks, cybersecurity, agentic safety, …
This document is Anthropic's internal policy for deciding when its AI systems are safe enough to keep developing and deploying. It sets specific capability thresholds that trigger safety requirements — …
These are the rules governing your use of Anthropic's consumer Services. Anthropic can change, suspend, or cut off your access at any time without warning, and its financial liability to …
Anthropic's Acceptable Use Policy sets out what users across all product surfaces (Claude.ai, API, cloud integrations) may and may not do with Claude. The policy's most operationally significant provisions establish …
This document establishes the commercial terms governing business and developer access to Anthropic's Claude API, excluding individual Claude.ai users who are covered under separate terms. The agreement specifies that customers …
This policy describes how Anthropic collects and uses your data, including sensitive information like facial images, and explains that your conversations with Claude may be used to train Anthropic's AI …
Anthropic updated its Usage Policy effective November 12, 2026, restructuring how it categorizes and enforces acceptable use rules. The policy now explicitly divides requirements into three categories (Universal Standards, High-Risk …
View change record →Anthropic updated its Privacy Policy effective date from July 8, 2026 to September 10, 2026 and substantially restructured its Korea-specific disclosures. The prior Korea section listed a domestic representative and …
View change record →Anthropic updated its privacy policy effective July 8, 2026, restructuring how it describes data collection and use. The policy now explicitly discloses new categories of personal data collection, including biometric …
View change record →Get the weekly coverage letter
Companies change their terms without announcing it. We capture governance documents, compare each new version with the last, and publish what changed, with the source text behind it. One email a week on what we found and what we could not reach.
Anthropic is more transparent than most AI companies about data retention. Here's exactly what happens when you delete your data, and how t…
ConductAtlas tracks 8 Anthropic documents including terms of service, privacy policy, and other governance documents. Every document is captured daily with cryptographic verification.
Anthropic has made 3 policy changes in the past 12 months across the documents ConductAtlas tracks.
ConductAtlas has classified 21 provisions across Anthropic's tracked documents.
Yes. Monitor subscribers ($4.99/month) can add Anthropic to their watchlist and receive an email alert after the daily check that finds a change to a tracked document. We check daily where we can, and show you where we can't.