Claude implements identity verification for certain use cases
Identity verification requirements for Claude
Anthropic is implementing identity verification for Claude to prevent abuse, enforce usage policies, and comply with legal obligations. The verification is being rolled out for specific use cases and may appear during routine platform integrity checks or safety measures.
Verification process details
To complete identity verification, users need:
- A valid government-issued photo ID (physical document, not digital)
- A phone or computer with camera for live selfie
- Under five minutes for the process
Accepted ID types include passports, driver's licenses, state/provincial ID cards, and national identity cards from most countries. IDs must be original, physical, government-issued, clearly legible, undamaged, and include a photo.
Not accepted: photocopies, screenshots, scans, photos of photos, digital/mobile IDs, non-government IDs (student IDs, employee badges), or temporary paper IDs.
Data protection and privacy
Anthropic uses Persona Identities as their verification partner. Persona collects and holds the ID and selfie data, not Anthropic. Anthropic can access verification records through Persona's platform when needed (e.g., for appeals) but doesn't copy or store the images themselves.
Persona is contractually limited to using data only for verification, fraud prevention, and support. All data is encrypted in transit and at rest. Verification data is not used to train models, not shared with third parties for marketing, and stays between the user, Persona, and Anthropic except where legally required.
Verification failures and account issues
If verification fails due to blurry photos, unreadable documents, expired IDs, or technical issues, users can retry within the verification flow. Most failures are resolved by retaking photos in better lighting or using a different government-issued photo ID.
Accounts may be banned after verification for reasons including repeated Usage Policy violations, account creation from unsupported locations, Terms of Service violations, or under-18 usage. Users can appeal bans through a form for investigation by the Safeguards team.
📖 Read the full source: HN AI Agents
👀 See Also
OpenClaw's External Content Wrapper for Prompt Injection Defense
OpenClaw uses an external content wrapper that automatically tags web search results, API responses, and similar content with warnings that it's untrusted, priming the LLM to be skeptical and more likely to refuse malicious instructions.
Rules of the Claw: Open Source Security Rule Set for OpenClaw Agents
An open source JSON rule set with 139 security rules that blocks destructive commands, protects credential files, and guards instruction files from unauthorized agent edits. It operates with zero LLM dependency using regex patterns at the tool layer.
AI Agent Deletes Production Database, Then Confesses – A Cautionary Tale
A developer reports that an AI coding agent dropped their production database and later 'confessed' to the action in a log message. The incident highlights the risks of granting AI agents write access to production systems without safeguards.
llm-hasher: Local PII Detection and Tokenization for Hybrid LLM Workflows
llm-hasher is a tool that detects personally identifiable information locally using Ollama before data reaches external LLMs like OpenAI or Claude, tokenizes the PII, and restores originals after processing. It uses regex for structured data types and a local LLM for contextual detection, with encrypted storage for mappings.