Anthropic doubles down on safety with Claude ID checks | Official Website of Louis Velazquez, entrepreneur, finance guy and tech innovator

The internet spent two decades teaching people they could be anyone. You picked a username, typed in an email, and the service took your word for it.

That quiet bargain is expiring now, one login screen at a time, and the companies building the most powerful software on earth are leading the change.

Artificial intelligence (AI) chatbots grew up inside that loose arrangement. You signed up, clicked past a few terms, and started typing. Anthropic, the company behind the Claude assistant, has long told users they must be at least 18, though for most people that rule lived in the fine print rather than at the front door.

Over the past year, a wave of state and national age-verification laws has started forcing tech platforms to actually check, not just ask.

That front door is now changing for Claude. Under a new section of its privacy policy, published in June and set to take effect July 8, Anthropic says it may ask certain users to prove their age or identity “in certain circumstances,” according to TechCrunch.

In practice, that can mean uploading a scan of a government passport or driver’s license and sitting for a facial scan.

The pressure on Anthropic did not start with ID checks

To understand why this is important, look past the privacy policy itself.

Anthropic is not a small startup hoping to seem responsible. It is one of the most valuable private companies on the planet, worth roughly $965 billion after its most recent funding round, TechCrunch noted.

More AI:

It is also in a fight with its own government. Over the past four months Anthropic has been labeled a national security risk, had a top model forced offline, and ended up negotiating with the White House over who is even allowed to use its tools, according to my TheStreet coverage.

President Donald Trump has directed federal agencies to stop using the company’s products, turning a regulatory dispute into a commercial one.

Anthropic also confidentially filed for an initial public offering (IPO) on June 1, a debut that could land this fall, according to TheStreet.

That timing changes how every policy move reads. A company weeks away from selling stock to the public has strong reasons to show regulators it knows exactly who is using its product.

Anthropic’s updated privacy policy can prompt some Claude users to verify their age or identity.

Supatman / Getty Images

What an identity check on Claude actually involves

Here is what the policy describes. When a check is triggered, a user can be asked to upload a scan of a government passport or driver’s license, record a selfie photo or video, and submit to a face-geometry scan, according to TechCrunch.

Illinois treats that kind of facial data as legally protected biometric information.

Anthropic says the change is narrower than it looks. The update applies only to a “small subset of users” whose accounts were flagged but not banned, and works as a way to appeal that flag, according to an X post from Anthropic’s Thariq Shihipar.

The company says the policy is unrelated to its newest models.

The policy language itself reads wider. Anthropic says users may see a verification prompt “as part of our routine platform integrity checks,” according to the company, and it lists fraud prevention, terms enforcement, and security investigations among its reasons.

Those are not the same thing as a one-off appeal.

The checks run through Persona, a San Francisco identity-verification vendor, according to Anthropic. Persona is backed by Founders Fund, the firm started by Trump backer Peter Thiel, who also holds a stake in Anthropic.

Discord selected Persona for age checks earlier this year, then backed off after user pushback, TechCrunch confirmed.

Anthropic decides how long Persona keeps the documents, but its spokesperson did not say when this information is deleted, according to TechCrunch. For comparison, Roblox (RBLX), another Persona client, says it deletes user images “immediately” after processing.

Anthropic has not spelled out exactly what trips a flag, and that gap is the part privacy researchers tend to fixate on.

A face-geometry template is not a password you can reset. Once a vendor holds a digital map of your face, a future breach or a government subpoena reaches data you can never change.

What the Claude ID rule means for AI investors

Most readers cannot buy Anthropic stock. The company is private, so the practical exposure runs through the public names that back it or supply it.

Amazon (AMZN) is one of Anthropic’s largest investors and its main cloud host. Alphabet (GOOGL) has also poured money in. Nvidia (NVDA), which holds stakes across the AI field, trades as the closest liquid proxy for the whole bet.

If you own an S&P 500 index fund, you already hold a slice of this story through Amazon and Alphabet. Neither stock moves on a single privacy tweak. Both move on whether Washington lets Anthropic operate normally, and that is the harder question sitting underneath this one.

When I lined up this ID change against the past four months, a pattern emerged. A company under this much government pressure is being pushed to track its users more tightly, and that instinct rarely stays contained.

The same reflex that asks a flagged user for a passport is the one a future regulator can lean on.

Here is how the standoff escalated, based on prior reporting.

The Pentagon labeled Anthropic a supply-chain risk earlier this year, according to TechCrunch.
Anthropic sued the Pentagon over that label in March, according to federal court records.
The Commerce Department forced two of Anthropic’s top models offline on June 12, my TheStreet report highlights.
The White House and Anthropic began drafting a shared AI risk framework on June 18, according to my TheStreet coverage.

My read is that the ID policy is a small line item in a much larger negotiation over control. An IPO at a roughly $965 billion valuation would drop Anthropic exposure into pensions and retirement accounts that already ride the AI trade, and underwriters will want the identity and data-handling language airtight before that day.

For investors, the takeaway has little to do with one upload screen. The real signal is that the AI names propping up those accounts now answer to a government willing to reach inside a product and rewrite how it works.

So the catch waiting for some Claude users is real, and it doubles as a preview.

The next time a chatbot asks you to prove who you are, the request will feel routine instead of strange, which is the part worth noticing. When Anthropic’s IPO paperwork becomes public this fall, the section on data and identity is where this quiet change begins carrying a price tag investors can read.