Skip to content

Usage Policy

20261021_rev01 → 20261022_rev01COLOUR MARKS THE SEVERITY OF A FLAGGED CLAUSE · + AND − MARK ADDED AND REMOVED
1Our Usage Policy (also referred to as our "Acceptable Use Policy" or "AUP") applies to anyone who submits inputs to Anthropic's products or services, all of whom we refer to as "users." This includes individuals using our apps (such as [Claude.ai](https://claude.ai/) and Claude Code), developers and businesses using our API and developer platforms, customers accessing Claude through cloud providers and authorized resellers, and the end users of products or services integrating Claude. The Usage Policy is intended to prevent our products and services from being used to create or contribute to real-world harm, and to promote the responsible use of our products and services.
2
3**Please review this Usage Policy carefully before using our products or services.**
4
5We will update our policy as our technology and the associated risks evolve or as we learn about unanticipated risks.
6
7Our Usage Policy is divided into three categories:
8
9- **Universal Usage Standards:** Our Universal Usage Standards are prohibitions that apply to all users and use cases.
10- **High-risk Use Case Requirements:** Our High-risk Use Case Requirements apply to specific use cases that pose an elevated risk of harm to individuals.
11- **Additional Use Case Guidelines:** Our Additional Use Case Guidelines apply to certain other use cases, including consumer-facing chatbots and agents, products serving minors, agentic use, and software listed in our directories.
12
13Our Usage Policy focuses on _how_ our products and services are used. Its prohibitions and requirements do not categorically prohibit use by any particular industry or line of business, provided the use of our products and services complies with this Usage Policy.
14
15Anthropic's Safeguards Team implements detection and monitoring to enforce our Usage Policy. If we suspect that you may have violated our Usage Policy, we may warn you or throttle, limit, suspend, or terminate your access to our products and services.
16
17Our products also include real-time safeguards that may block or limit certain outputs, but encountering a block does not by itself indicate a violation of the Usage Policy. For certain dual-use or high-risk use cases covered by real-time safeguards in select domains, we may offer programs where you can apply to be granted access to our products and services with adjusted safeguards, such as our [Cyber Verification Program](https://support.claude.com/en/articles/14604842-real-time-cyber-safeguards-on-claude-opus-and-sonnet) and [Life Sciences Verification Program](https://claude.com/form/life-sciences-verification-program), subject to program-specific eligibility criteria.
18
19If you believe that our model outputs are potentially inaccurate, biased, or harmful, please notify us at usersafety@anthropic.com, or report it directly in our product through the "report issues" thumbs down button or similar feedback features (where available). You can read more about our Safeguards practices and recommendations in our [Safeguards Support Center](https://support.anthropic.com/en/collections/4078535-safeguards).
20
21_This Usage Policy is calibrated to strike an optimal balance between enabling beneficial uses and mitigating potential harms. Anthropic may enter into contracts with certain governmental customers that tailor use restrictions to that customer's public mission and legal authorities if, in Anthropic's judgment, the contractual use restrictions and applicable safeguards are adequate to mitigate the potential harms addressed by this Usage Policy._
22
23## Universal Usage Standards
24
125### Do Not Violate Applicable Laws or Engage in Illegal Activity
226
327This includes using our products or services to:
428
529- Illegally produce, acquire, sell, or distribute controlled or regulated substances
630- Engage in or facilitate human trafficking or prostitution
163187This includes using our products or services to:
164188
165189- Depict or request sexual intercourse or sex acts
166190- Generate content related to sexual fetishes or fantasies
167191- Facilitate, promote, or depict incest or bestiality
168192- Engage in erotic chats
193
194## High-risk Use Case Requirements
195
196Certain uses of our products and services can impact an individual's health, legal rights, finances, livelihoods, or access to housing, education, or essential services. When our products are used for these high-risk use cases, we require that a qualified person remain accountable for what is delivered or decided, and that the impacted individual knows AI was involved in the advice or decision. These requirements apply to anyone who uses our products to make a High-risk AI Recommendation or control equipment capable of High-risk Physical Actions.
197
198_These requirements for high-risk use cases are meant to protect individuals. However, they do not replace local legal requirements and restrictions. Users should always default to following local laws if those laws conflict with or add additional requirements to the requirements listed below. Our products and services may not be suitable for all use cases._
199
200### Requirements
201
202For High-risk AI Recommendations or Physical Actions, users must implement the following controls:
203
204- **Qualified human-in-the-loop** **for advice and decisions** - A qualified person must meaningfully review a High-risk AI Recommendation, with the authority to change or modify it, before it is delivered as advice to an individual who might rely on it or before it is used to implement a decision. A qualified person is someone with the training or experience to evaluate the output in that field, and who holds a license where applicable law requires one for the underlying activity. The qualified person remains responsible for the accuracy and appropriateness of what is delivered or decided.
205- **Disclosure for advice and decisions** - An individual who receives advice or who is the subject of a decision based on a High-risk AI Recommendation must be clearly told that AI was used to produce it. Users do not need to identify Anthropic, Claude, or the model used.
206- **Human control and independent safety limits for physical actions** - For High-risk Physical Actions, there must be a qualified individual who is able to observe the equipment's operation and stop it at any time, and the equipment must stop or hold a safe state when that individual intervenes or when connection to our services is lost. Any operating limits to keep the equipment within safe bounds, such as speed, force, reach, temperature, pressure, voltage, energy output, dose, or operating area, must be enforced by the equipment or a controller independent of model output.
207
208These requirements apply alongside applicable law. For physical devices, they do not replace vehicle, aviation, medical device, machinery, or workplace safety law or certification.
209
210### High-risk AI Recommendations
211
212The requirements apply when users use our products or services to provide a recommendation in any of the following, which are "High-risk Areas":
213
2141. **Legal** - advising an individual on their legal rights, obligations, or liability, or on what to do in a specific legal matter, or preparing legal documents for them to sign or file.
2152. **Medical** - diagnosing a condition; recommending, starting, stopping, or adjusting a treatment or medication, including dose; interpreting an individual's symptoms or test results to triage, provide a diagnosis, advise on treatment risks or side effects, or create a treatment plan. This includes assessing, diagnosing, or treating a mental health condition.
2163. **Finance** - a personalized recommendation to buy, sell, hold, or allocate among specific investment products, such as securities, funds, digital assets, or annuities; advising an individual on their own taxes, such as what to report, deduct, or claim; or preparing a tax return for them to sign or file.
2174. **Credit** - approving, denying, or setting the amount, rate, or limit of a loan, mortgage, credit line, or similar product, or reducing or closing an existing one, or producing a credit score or creditworthiness assessment used for such decisions.
2185. **Insurance** - underwriting or pricing a policy; determining the outcome of a claim; making a coverage, prior-authorization, or medical-necessity determination; cancelling or declining to renew a policy.
2196. **Housing** - approving, denying, or ranking a rental or purchase application; terminating a tenancy or initiating eviction or foreclosure.
2207. **Employment** - screening, ranking, advancing, or rejecting candidates; hiring; promoting or demoting; setting or changing shifts, hours, or pay; evaluating a worker's performance or behavior, including through disciplining, suspending, or terminating a worker.
2218. **Education and credentials** - admitting, rejecting, or ranking applicants; grading an exam or assessment that determines admission, graduation, licensure, or certification (outside of multiple-choice testing); granting, denying, or revoking a professional license or certification; making a determination of academic misconduct or suspending or expelling a student.
2229. **Healthcare access** - setting triage or waitlist priority; determining eligibility for, denying, or discontinuing a treatment, service, or care program.
22310. **Public benefits and services** - determining eligibility for, or the amount of, government benefits, assistance, or essential services, including suspension and termination.
22411. **Legal status and adjudication** - deciding or recommending the outcome of a civil, administrative, or arbitral matter; determining immigration, asylum, or citizenship status.
225
226**Prohibition Reminder:** Use of our products or services to make or suggest decisions in law enforcement or criminal justice processes is prohibited under our Universal Usage Standards and is not a High-risk AI Recommendation.
227
228### High-risk Physical Actions
229
230When connecting our models to hardware that can take physical actions, users should assess and understand 1) the range of possible outcomes if something goes wrong, 2) the nature and extent of potential impact to individuals, groups or populations, and 3) the degree of control the user maintains relative to the autonomy the model has, including possible risks such as prompt injection or manipulation. The requirements apply when outputs of our products and services are acted upon by hardware without human approval where the hardware or linked equipment can:
231
2321. **Move through shared space** - steer, accelerate, brake, or navigate a vehicle, vessel, aircraft, drone, or mobile robot in a place where people may be present, or on roads, waterways, or airspace.
2332. **Apply force that could cause injury** - actuate arms, presses, lifts, hoists, doors, gates, conveyors, or other machinery capable of striking, crushing, pinning, or dropping a load on a person nearby.
2343. **Control hazardous energy or materials** - operate heating elements or open flame, pressure systems, high-voltage or high-current circuits, lasers or radiation sources above eye- and skin-safe levels, or control the release, mixing, or dispensing of hazardous chemical or biological materials, at levels capable of causing injury.
2354. **Act on the human body** - administer or adjust a drug, fluid, gas, electrical stimulation, or radiation or other energy dose to a person, or cut, penetrate, or move instruments within a person's body or tissue, including through a surgical, interventional, worn, attached, or implanted device.
2365. **Control safety systems** - arm, disarm, trigger, or override fire suppression, emergency shutoffs, alarms, ventilation, or locks and barriers that could confine a person.
2376. **Run industrial processes** - control plant or process equipment where a fault could injure workers or a member of the public (including diffuse or indirect harms, such as food poisoning or defective products).
238
239**Prohibition Reminder:** Our Universal Usage Standards prohibit actions related to weapons, critical infrastructure, violence, and surveillance.
240
241### Exclusions
242
243The requirements do not apply to:
244
245#### _AI Recommendations_
246
247Using our products to assist in workflows that are related to, but do not actually provide, High-risk AI Recommendations. Some non-exhaustive examples are:
248
249- General information or educational content, including explaining what a law says, how a condition or treatment works, or how investing works, without applying it to the individual's circumstances.
250- Wellness content, such as education on sleep, stress, nutrition, or exercise.
251- Explaining or helping an individual understand advice, a diagnosis, or a decision that a qualified person has already given them (as long as it is not an additional High-risk AI Recommendation).
252- Internal drafting, research, summarization, or analysis that is not the ultimate High-risk AI Recommendation delivered.
253- Carrying out a decision a person has already made, or applying a fixed rule, formula, or algorithm where the model exercises no judgment about the individual.
254- Business operations that do not advise or decide about a specific individual.
255
256#### _Physical Actions_
257
258- Consumer and household devices operating within the manufacturer's built-in safety limits, where the worst possible outcome is discomfort or inconvenience.
259- Generating plans, code, toolpaths, waypoints, or commands that a qualified person reviews before they are executed on equipment.
260- Monitoring, sensing, or reporting on equipment without controlling it.
261- Laboratory automation handling non-hazardous materials inside enclosed instruments.
262
263### Favorable Decisions
264
265Where permitted by law, human-in-the-loop review is not required before acting on a High-risk AI Recommendation wholly favorable to the individual in the following circumstances:
266
267- Paying an insurance claim.
268- Approving coverage or prior authorization for care a clinician has ordered.
269- Granting eligibility for a healthcare service or program.
270- Granting or continuing a public benefit or service.
271
272Additional legal obligations such as disclosure may still apply. A partial approval, reduced amount, or approval with conditions is not considered "wholly favorable."
273
274## Additional Use Case Guidelines
275
276The below use cases - regardless of whether they involve High-risk AI Recommendations or Physical Actions - must comply with the additional guidelines provided.
277
278- **All consumer-facing chatbots or AI agents that communicate directly with external users** must **disclose** to users that they are interacting with AI rather than a human. This disclosure must be provided at a minimum at the beginning of each chat session or in the product interface. The disclosure does not need to identify Anthropic, Claude, or the model used.
279- **Products serving minors,** including organizations providing minors with the ability to directly interact with products that incorporate our API(s), must comply with the **additional guidelines** outlined in our Support [article](https://support.anthropic.com/en/articles/9307344-responsible-use-of-anthropic-s-models-guidelines-for-organizations-serving-minors) and all applicable laws.
280- **Agentic use cases** must comply with the Usage Policy. Users are responsible for ensuring that agents they build or deploy, including actions those agents take through tools, browsers, or connected systems, comply with this Usage Policy.
281- **Software** listed in our Directories must comply with our [Software Directory Policy](https://support.claude.com/en/articles/13145358-anthropic-software-directory-policy).
282
283We also provide additional informational resources in our [Safeguards Support Center](https://support.claude.com/en/collections/4078535-safeguards), including:
284
285- Our [Child Safety Guidance for Developers](https://support.claude.com/en/articles/15591275-child-safety-guidance-for-developers) for detailed guidance on identifying and mitigating child safety risks.
286- Additional information about our [mental health safeguards](https://support.claude.com/en/articles/13171706-crisis-helpline-support-in-claude) on Claude.ai.