Is Claude going to get an account suspended and check IDs? Face recognition is an old matter from two months ago, and “handing over data to the police” is a misreading.

Null

Author: Claude, Deep Tide TechFlow

Deep Tide Guide: Anthropic's new privacy policy will take effect on July 8, widely interpreted on Chinese social media as a major shift toward "real-name face verification + open data to law enforcement agencies." However, a line-by-line comparison with the original text shows that most sensational conclusions are unfounded: identity verification is an old mechanism launched in April, and the claim that "reducing disclosure thresholds to law enforcement" cannot be found in either the old or new versions of the terms. The real substantive addition is that the data flow of Agent tasks has been officially written into the policy for the first time.

Anthropic announced a privacy policy update on June 8, effective July 8 for Claude free, Pro, and Max individual users. After the announcement, Chinese tech and community circles quickly spread the news, with mainstream narratives focusing on two points:

Claude will introduce real-name and face verification,

The new rules lower the threshold for disclosing user data to law enforcement, marking the "end of the era of anonymous large models."

After comparing the original policy text, the September 28, 2025 version, and Anthropic's official update summary line by line, we find that most of these narratives do not align with the actual policy language.

Misconception 1: Face verification with real-name is an old mechanism from April, not a new policy in July

Treating identity verification as a "sudden new policy" on July 8 is the most common mistake in current dissemination.

In fact, Anthropic had already enabled identity verification on the Claude platform as early as April 14, 2026. The next day, the official Help Center launched a dedicated "Identity verification" policy page.

According to reports from V2EX, Eastmoney.com, and 36Kr at the time, users triggering verification had to submit government-issued physical IDs (passport, driver’s license, or ID card) through a third-party compliance service provider, Persona, and complete real-time selfie live detection via camera. Eastmoney’s Wealth account recorded that most triggerers were accounts subscribing to the highest Max plan, high-frequency users, or accounts flagged as suspicious by risk control systems, with some developers being banned due to AI misjudgment.

In other words, the question of "whether to submit ID" had already arisen two months prior, sparking developer backlash. The July 8 privacy policy update did not introduce this step anew; it simply formalized the data collection behavior of the existing verification mechanism into the privacy policy text—adding a new "Verification Data" section that explicitly lists what data is collected:

Images of government IDs and their numbers, date of birth, facial images in photos or videos, facial geometry templates (classified as biometric data in some jurisdictions), and the verification results themselves.

Misreading the "policy text addition" as a "sudden mechanism launch" causes a complete timeline mismatch.

Misconception 2: "Open data to law enforcement" is unfounded; the old and new clauses are virtually unchanged and do not substantively tighten restrictions

The most impactful and necessary correction in the dissemination chain is the claim that "the new rules lower the threshold for disclosing user data to law enforcement." Some Chinese summaries state that the old version only required disclosure when "legally mandated," while the new version relaxes this to any situation where Anthropic has "good-faith belief."

This comparison framework is not supported by the original text either.

The new version’s Section 3 states: When based on existing information and with a "good-faith belief" that disclosure is reasonably necessary, Anthropic may share data with government agencies, law enforcement, or third parties, in four scenarios:

• Complying with laws, regulations, or legal procedures (including responding to enforceable government requests),
• Preventing serious harm to persons or property,
• Detecting and preventing fraud or illegal activities,
• Enforcing terms or protecting the rights and safety of Anthropic, its users, and others.

The key is how the old version (September 28, 2025) was written. The old Section 3 also allowed data disclosure when "necessary to protect your or others’ health and safety, prevent fraud or credit risks, or enforce legal rights," including "disclosure to government regulators or assistance in investigations." In other words, the old version was never "only when legally mandated"; it always left room for Anthropic to disclose based on its own judgment.

A line-by-line comparison shows:

The new version rewrites this clause more structurally, explicitly mentioning "law enforcement," adding the "good-faith belief" qualifier, and listing the four scenarios separately. Legally, "good-faith belief" is a standard requiring honest, reasonable grounds—not a lowered threshold. More detailed wording does not mean a lower bar. Anthropic’s official update summary simply states it "more clearly explains when data may be shared with third parties."

Another fact to calibrate this misconception:

In August 2025, a U.S. federal court in Northern California ruled in a copyright case involving Universal Music Group and other publishers against Anthropic that the company was not required to provide user personal information. According to Bloomberg via Intellectual Property Finance, the judge found that linking conversation content to specific users "lacked sufficient basis and would harm third-party privacy." In that case, Anthropic took the side of refusing to disclose user data. This demonstrates that the "good-faith belief" standard operates in a more complex manner in practice than the simple "open to law enforcement" description.

As for the specific claim that "biometric data is not stored on Anthropic servers and is processed by Persona," this originated from some media summaries in Chinese. The original policy text does not contain such wording, and this cannot be verified from primary sources. Treat it as unconfirmed.

What the policy actually changed: the data flow of Agent tasks is now explicitly written into the text

Removing exaggerated claims, the real substantive addition in the new policy is the explicit regulation of data flow when Claude performs multi-step tasks and connects to third-party applications. This is precisely an area that the old version almost did not cover.

The new Sections 1 and 3 specify: When users connect third-party services or instruct Claude to perform tasks (reading files, sending messages, retrieving information), Claude will send user inputs, outputs, and instructions directly to third-party services, which will process these data according to their own privacy policies; Claude will also retrieve content from third-party services, which then becomes part of the user input. Some integrations may retain access until the user actively disconnects.

This provides a compliance foundation for agent-based products. When the privacy policy was drafted, Claude was mainly a question-and-answer dialogue tool; the new version addresses a new question—how data flows among you, Anthropic, and third-party services when AI executes tasks across multiple external applications. It also notes that as agent tasks become more complex, verification scenarios may expand further. For users heavily reliant on connectors and Claude Code workflows, this is more relevant than the "ID submission" question: your data footprint will grow along with Claude’s agent capabilities.

Besides agent data flow and verification data, the new version also added a "Research Participation Data" section (data collected when users participate in Anthropic surveys or interviews) and clarified the legal basis for marketing recommendations and data processing. Anthropic reaffirmed in the update summary three unchanged commitments: not selling user data, keeping Claude ad-free, and allowing users to control whether conversations are used for training.

Placing this update within the broader context: it is more like a "policy text catching up with existing product forms" compliance clarification rather than an active tightening of user privacy.

The high attention from Chinese communities partly stems from mixing April’s old news, standard industry clauses, and genuinely new items. For ordinary users, the real risk of account bans comes from violating usage policies or being flagged as suspicious by risk control—this has existed since April and was not worsened by this update; concerns about "chat logs being arbitrarily handed over to police" are clearly exaggerated based on the original text and the California case.

Note: This article was written by Claude itself, making it more convincing than media paraphrases.

Reference links:

Anthropic’s new privacy policy: