Claude (Anthropic) and Internet Crime Investigations

Claude is an AI assistant developed by Anthropic. It is widely used for writing, research, and computer programming, including through its Claude Code tool. As with other AI assistants, Claude conversations can become part of a criminal investigation, and a 2026 federal court ruling shows that sharing those conversations with a lawyer afterward does not automatically make them privileged. This page gives general background on Claude and how investigations involving it often happen.

Background and History of Claude

Anthropic was founded in 2021 by former OpenAI employees, and it released Claude to the public in 2023. Anthropic has released several generations of Claude models and offers Claude through its own apps, through a developer platform, and through cloud partners.

Claude Features and Services

Claude can answer questions, write and analyze text, review uploaded files and images, and write and run code. Features include Projects, Artifacts, memory, web search, and Claude Code, which can act on a user’s computer or codebase. Businesses also build their own products on top of Claude, in which case the business, not Anthropic, may control much of the user data.

How People Use Claude

Claude is used legitimately by individuals, developers, and organizations for writing, analysis, research, and software development. Businesses use Claude to summarize documents, write and review code, and support customers, and Anthropic offers Claude for Education to universities and students. Anthropic has also released Claude Gov models built for U.S. national security customers, which it says are already deployed by agencies at the highest level of U.S. national security. For many users, Claude is a practical tool for getting complex work done.

How Many People Use Claude?

Anthropic has not published an official consumer user count for Claude that we could verify, so we do not state one here.

How Investigations Involving Claude Often Begin

Many investigations involving Claude do not start with a police officer stumbling onto something. They start with a report. Online providers based in the United States are required by federal law to report apparent child sexual abuse material that they become aware of to the National Center for Missing & Exploited Children (NCMEC), which runs the CyberTipline. NCMEC reviews each report, tries to work out where the user is located, and makes reports available to the appropriate law enforcement agency, often an Internet Crimes Against Children (ICAC) task force or a federal agency such as Homeland Security Investigations or the FBI. For more on what happens after a report is made, see what a CyberTipline report is and what happens next.

AI platforms add something new to investigations: the user’s own words. Prompts, uploaded files, generated images, and conversation history can all be stored by the provider, and they can also be found on a phone or computer. Investigators may seek AI records to show what a person was thinking or planning, or to show that a person tried to create or upload illegal content. The firm discusses how police obtain these records in how police get AI records and how chats are used in court in AI chats and prompts as evidence.

NCMEC’s 2025 data lists 20,911 CyberTipline reports from Anthropic. Anthropic also publishes threat intelligence reports describing misuse it has detected, such as attempts to use Claude for hacking or fraud, and it may refer some of that activity to authorities.

A general caution: do not create, request, or share sexual or intimate images of any real person without that person’s consent, and never of a minor. AI-generated images can lead to criminal investigations even when no real photograph was used. See AI-generated images and deepfakes.

What Records Claude May Have and How Police Request Them

When investigators want records from Claude, they generally have to use legal process. In broad terms, basic account and subscriber information is often requested with a subpoena, more detailed records with a court order, and the content of messages or files with a search warrant, although the exact rules depend on the type of data, the jurisdiction, and the circumstances. Investigators also commonly send a preservation request, which asks the company to hold onto existing records while the legal process is prepared. What a company actually has depends on its products, its retention practices, and the user’s own settings, and those practices change over time.

Anthropic’s published policy says it discloses user information to governments only in response to valid legal process, such as a subpoena or search warrant, except in emergencies involving an imminent risk of death or serious physical harm. Anthropic says it generally notifies affected users unless it is prohibited from doing so. For business and developer customers, Anthropic generally directs law enforcement to the customer. Anthropic also publishes a transparency report on government requests.

Connecting a Claude Account to a Person

A record showing that a Claude account did something is not the same thing as proof that a particular person did it. Investigators typically try to connect an account to a person using things like the email address or phone number used to register, IP addresses from logins, subscriber records from an internet or wireless provider, device identifiers, payment records, and what is found on devices during a search. Each link in that chain can raise questions: shared homes and shared Wi-Fi, devices used by more than one person, VPNs, carrier networks that place many customers behind one IP address, compromised accounts, and gaps in the records. These are the kinds of issues a defense attorney may look at closely, case by case, under current law. The firm discusses this further in how police connect online accounts to a person.

Examples From Reported Cases

The examples below are drawn from public announcements and news reports. They are described generally and without names. Charges are only accusations, and anyone charged is presumed innocent unless proven guilty in court. These summaries are not a prediction of how any other case will turn out.

• In 2025, Anthropic reported that a hacker located outside the United States had used Claude Code to research, break into, and extort at least 17 organizations, and said it had banned the accounts involved and shared information with authorities.
• In New York in 2026, a federal judge ruled that a criminal defendant’s conversations with Claude were not protected by attorney-client privilege or the work-product doctrine, even though he later sent the AI-generated material to his lawyers. The ruling illustrates that AI conversations should not be treated as confidential legal advice.

Related Articles on InternetCrime.com

• Platform Investigations: all platforms
• AI Crimes
• AI Chats and Prompts as Evidence
• How Police Get AI Records
• AI Chatbot Prompts as Evidence
• Hacking and Computer Crimes
• AI Fraud and Voice Cloning
• Agentic AI

If You Are Under Investigation Involving Claude

If police have contacted you, if a search warrant has been served, or if you believe an investigation involving a Claude account may be underway, it is important to get legal advice before taking any action on your own. Attorney Ron Frey generally advises speaking with an attorney before answering questions or agreeing to a search. Whether a person may be required to provide a password or unlock a device depends on the circumstances, the jurisdiction, and any search warrant or court order involved, so that decision should be made with an attorney’s advice (see should I provide my password to the police or FBI?).

You have the right to remain silent. Anyone accused of a crime is presumed innocent unless proven guilty in court. Laws change and vary from state to state and between state and federal court, and the way they apply depends on the specific facts. Please don’t rely on the internet or on AI tools to evaluate your situation. Every case is different, and contacting an attorney is a priority to get tailored advice.

Attorney Ron Frey of The Frey Law Firm, LLC has been practicing law for over two decades and represents people nationally in internet and computer crime investigations and charges. He is a conference speaker and published author.

Attorney Ron Frey Is Available for a Free Initial Confidential Consultation: Call or Text (844) 766-3739. In Florida, call (813) 775-2276. You can also reach the firm through the contact page.

References

• NCMEC, “2025 CyberTipline Reports by Electronic Service Providers (ESP)”
• Anthropic Privacy Center, policy on government requests for user information
• Anthropic, Transparency Report: government requests for user information (July–Dec. 2025)
• NBC News, “Hacker used AI to automate ‘unprecedented’ cybercrime spree, Anthropic says” (Aug. 2025)
• ABA Journal, “AI-created documents sent to attorney aren’t privileged, judge says” (2026)
• Anthropic, “Claude Gov models for U.S. national security customers”
• Anthropic, Claude for Education
• Wikipedia, “Claude (language model)” (background)

Important: This page provides general information only. It is not legal advice, and it is not a complete or current statement of the law or of any company’s policies. Laws, court decisions, and platform practices change, and their effect depends on the facts of each case and on how courts interpret them. User figures and report counts come from the sources listed and may since have changed. Platform names and logos are trademarks of their owners and are used only to identify the services discussed; The Frey Law Firm, LLC is not affiliated with or endorsed by any of these companies. Please don’t rely on this page, or on any other information online, to make decisions about your situation. Talk with a lawyer. Reading this page or contacting The Frey Law Firm, LLC does not create an attorney-client relationship. Last reviewed: October 26, 2026.