Anthropic’s Claude AI, long pitched as a safety-first alternative in the crowded chatbot arena, now demands proof of who you are. Some users log in and bam—prompted to whip out a passport or driver’s license for a live selfie check. No warning. No opt-out.
This isn’t blanket KYC. It’s targeted. Anthropic rolled it out quietly this week for “a few use cases,” per its Help Center. Think suspicious activity flagging fraud or abuse. Or accounts from blacklisted spots like mainland China, Russia, Iran. Underage slip-ups. Repeated policy breaks.
The trigger? Activity that pings Anthropic’s safeguards. “This applies to a small number of cases where we see activity that indicates potentially fraudulent or abusive behavior, which violates our usage policy,” an Anthropic spokesperson told Business Insider and Engadget.
Users grab a physical government ID—passport, driver’s license, national card from most countries. Hold it up. Snap a selfie via phone or webcam. Done in under five minutes, usually. Blurry shot? Expired doc? Try again. Fail too often, appeal via form.
Persona Identities handles the grunt work. San Francisco startup scans the ID, matches the face. Encrypts everything in transit and at rest. Anthropic calls the shots as data controller: sets retention, bans model training, limits sharing to legal musts. “We are not using your identity data to train our models. Verification data is used solely to confirm who you are,” the Help Center states flatly.
But. Persona doesn’t operate in a vacuum. Its privacy policy lists 17 subprocessors—AWS, Google, Stripe, even OpenAI. Data hops through that chain. Users fret. One X post blasts: “Anthropic making unexplainable decisions.” Another: “We are living in 1984.” Screenshots of the prompt flood feeds.
Persona’s Shadow: From Discord Debacle to AI Frontlines
Pick Persona, invite ghosts. The firm drew fire last year with Discord’s age-gate test. Allegations flew: data exposed on a government server. Discord paused. Persona denied. Now, back in the spotlight. Investors? Founders Fund, Peter Thiel’s outfit. Thiel chairs Palantir—FBI, CIA, ICE clients. Coincidence? Users connect dots to surveillance.
Anthropic picked them for “strength of their technology, privacy controls, and security safeguards.” Data lives on Persona’s systems. Anthropic peeks for appeals but doesn’t hoard images. Still, post-verification bans hit hard. Reasons: policy repeats, unsupported locales, ToS slips, kids under 18. Appeal if you dare.
No peers match this yet. OpenAI’s ChatGPT? Credit card suffices. Google’s Gemini? Same. Claude breaks ground as first big AI with government ID walls. Privacy hawks cry foul. Why hand passports to a chatbot firm when cards already prove you’re paying? Help Net Security warns of trust erosion.
In China, panic ripples. Claude’s coding chops draw devs despite bans. Relay platforms like AICodeMirror host thousands. Now? Black markets boom. Xianyu vendors hawk African proxies for 150 yuan ($22) a pop—”tens of thousands” ready, one claims. Shenzhen engineer Zhang, years deep on Singapore-masked access, shrugs: “Chinese ID won’t cut it.” Eyes GPT, Gemini. South China Morning Post tracks the scramble.
Abuse Fighters or Access Barriers?
Fraud’s real. Bans evade with sock puppets. Kids game rules. Sanctioned zones skirt via VPNs. Verification plugs holes. Enforces age gates—Claude paused a 15-year-old’s account, demanding 18+ proof, per The New Stack. But at what cost? Subs cancel. Devs pivot. X erupts: “You switched to Claude over surveillance fears. Now it wants your passport,” posts @MarsSignals.
Anthropic doubles down. “We ask for the minimum information required to verify your identity.” No marketing shares. No excess collection. Yet Persona’s Discord stink lingers, as The Register notes—subprocessors galore, retention vague. Users demand: delete my data post-check? Anthropic points to policy.
Broader strokes. AI firms chase responsibility badges amid regs like Europe’s AI Act. OpenAI uses Persona for age checks too. Roblox too. Verification spreads. Claude’s move signals more walls ahead. Devs in gray zones? Black markets thrive. Privacy purists? Flee to no-ID alts.
One dev gripes on X: “My passport number is now floating around services I’ve never heard of.” Spot on. Persona verifies against 200+ country databases, including China’s. Proxies may flop if behavior betrays.
Anthropic bets the trade-off pays. Fewer bots. Cleaner platform. But users vote with feet. Claude’s edge? Safety sold it. Now, that pitch frays. Watch subs. Watch churn. The ID gate tests loyalties.
WebProNews is an iEntry Publication