Identity verification on Claude

Scope and purpose of ID verification

• Anthropic is rolling out ID checks for “certain capabilities” and suspected under‑18 accounts; consumer plans are included, enterprise status is unclear.
• The support page and flow have existed since at least April, but the privacy policy was recently updated to allow broader age/identity checks, prompting renewed debate.
• Some speculate this will be used to gate access to Fable/Mythos-class models or satisfy US export‑control requirements; others note the page itself predates the Fable ban.

Use of Persona

• Many are unhappy that Anthropic chose Persona, citing earlier criticism, data‑leak reports, and connections to US surveillance‑adjacent firms and investors.
• Commenters highlight that Anthropic says it won’t train on ID data, but Persona may use it to “improve fraud detection,” i.e., to train its own systems.

Privacy, surveillance, and government access

• Strong concern that tying real‑world identity to detailed AI conversations creates a powerful surveillance and censorship tool, especially when US legal process can compel data.
• Several argue this formalizes a “no‑fly list for AI” or “bank‑style” deplatforming, enabling quiet discrimination or account “nerfing.”
• Others note that credit cards already reveal identity, so the marginal privacy loss is smaller; the main new risk is biometric data.

Impact on users and behavior

• Many say ID prompts would be an immediate trigger to cancel Claude subscriptions; some already have.
• Others will comply if required to regain Fable‑level access, treating it like any other KYC process.
• There is anxiety about opaque triggers (e.g., being misclassified as underage or “risky”) and weak appeal/support channels.

Alternatives: local and non‑US models

• A large subthread discusses moving to open‑weight and especially Chinese models (DeepSeek, GLM‑5.2, Qwen, MiMo, Mistral), often via intermediaries like OpenRouter.
• Some claim these are close to or better than Claude/Sonnet for many coding and writing tasks, though others dispute parity with Opus‑class models.
• Commenters see US export restrictions and ID regimes as accelerating non‑US competition and making US LLMs a supply‑chain and geopolitical risk.

Regulatory and “AI neutrality” framing

• Comparisons are drawn to net neutrality and natural monopolies: whether frontier LLMs are becoming essential infrastructure vs. just SaaS.
• Several foresee identity‑linked access spreading to other AI providers and possibly to the wider internet, driven by “safety,” child‑protection, and national‑security narratives.