Anthropic Claude
Based exclusively on public evidence • 20 criteria (Privacy + AI)
Last review: 10 Feb 2026
AI Trust Summary
- •In AI: it does not detail the specific algorithmic logic of how complex decisions are made, which can generate uncertainties for contractors.
- •In Basic Privacy: it does not provide a Data Processing Agreement (DPA) for customers, which can compromise the security of processed data.
Safer Alternatives
Higher-rated software in the same category
Compliances in AI (3)
AI criteria the company meets. Buy the Premium Analysis to see all 3 criteria.
- •Anthropic Claude
- •documents the use of Inputs and Outputs to train models, ensuring clarity on data processing.
- •provides an opt-out mechanism for AI training, allowing greater user control.
- •these practices strengthen due diligence regarding privacy and AI use.
AI data retention policy clearly documented
The policy states that users can delete individual conversations, which are immediately removed from the history and deleted from the back-end within 30 days.
AI training opt-out control available
Offers a specific opt-out mechanism for model training through account settings, allowing greater user control.
Policy on data use for AI training clearly stated
Explicitly states the use of Inputs and Outputs to train models and improve services, ensuring clarity on data processing.
Source: vendor public documents
Highlights in Privacy (3)
Most relevant criteria for this category. Buy the Premium Analysis to see all 3 criteria.
Data Processing Agreement (DPA) not available for customers
The policy mentions that Anthropic acts as a data processor, but does not provide a direct link or explicit details of a DPA.
Data controller and processor roles clearly defined
Clearly distinguishes the roles of Controller and Processor, defining the scope of policy application.
Data controller identity and contact clearly disclosed
Provides full legal identity and physical contact addresses, promoting transparency.
Source: vendor public documents
Critical Alerts
- •Decisões automatizadas por IA explicadas de forma compreensível: Importante para a transparência nas decisões automatizadas..
- •Funcionalidades de IA claramente identificadas com suas finalidades: Ajuda a entender como as interações com o sistema são processadas e utilizadas.
Conformance analysis (20)
Deletion period for AI prompts and responses clearly defined
Reference: ISO/IEC 42001 (8.2) + ISO/IEC 27701 (7.4.6)
Opt-out control for AI training available
Reference: ISO/IEC 42001 (8.3) + ISO/IEC 29100 + EU AI Act
Policy on the use of AI prompts and responses for AI training declared
Reference: ISO/IEC 42001 (8.2) + ISO/IEC 23894 + EU AI Act
Source: vendor public documents
Follow this company and access all 20 criteria
Track score changes, get alerts on policy updates, and view the full conformance analysis
Don't miss any update
Sign up to follow this company and track changes in privacy and AI scores
Why trust the AITS Index: Open Community Audit
Public transparency, peer review and open evidence trails — all verifiable by the community
Trust guarantees
Peer review
users, professionals and experts confirm or contest items online.
Public history
vendor and index changes are versioned and accessible.
Participate
Evidence, confirmations and contestations
participate in the collaborative validation of AITS criteria
Other AI Tools software
Dive into in-depth research and analysis of each player
Source: vendor public documents
Analyzed Sources
Public documents used in the audit of Anthropic Claude:
Evidence, confirmations and contestations
participate in the collaborative validation of AITS criteria
Scope & Limitations
TrustThis/AITS assessments are based exclusively on publicly available information, duly cited with date and URL, following the AITS methodology (privacy & AI transparency).
The content is indicative in nature, intended for screening and comparison, not replacing internal audits.
TrustThis/AITS does not perform invasive tests, does not access vendor technology environments and does not process customer personal data. Conclusions reflect only the vendor's public communication at the date of collection.
Source: vendor public documents