Acceptable use

Buy My Agent is operated by BUYMYAGENT TECHNOLOGIES INC., a Canadian corporation (“we”, “us”). This Acceptable Use Policy forms part of our Terms of Service and applies to everyone who uses the marketplace.

Prohibited

• Illegal activity, unlawful goods or services, sanctions evasion, or anything prohibited by Stripe or another required provider.
• Fraud, scams, deceptive earnings claims, phishing, credential theft, impersonation, fake documents, or misleading affiliation.
• Malware, destructive code, security circumvention, data exfiltration, or anything designed to facilitate cyber abuse.
• Any sexual content involving minors, sexual exploitation or abuse, grooming, trafficking, or non-consensual intimate content.
• Credible threats, facilitation of violence, targeted harassment, hate, stalking, or doxxing.
• Collecting, selling, or inferring personal information without a lawful basis, clear notice, and consent.
• Giving individualized medical, legal, financial, tax, or immigration advice, or otherwise acting as a substitute for a licensed professional where licensing or individualized judgment is required. (General, clearly-labelled educational information is not the same as individualized advice — but agents that diagnose, prescribe, prepare filings, or tell a specific user what to do in these domains are not permitted at launch.)
• Emergency, crisis, or safety-critical agents — including suicide, self-harm, or mental-health crisis support; emergency, disaster, or medical-emergency response; and anything a person could rely on in a situation where a wrong answer risks physical harm.
• Agents directed at, or primarily intended for, children (under the age of majority), or that knowingly collect information from children.
• High-impact automated decisions about employment, housing, credit, insurance, healthcare, education, policing, or legal rights without lawful safeguards and human oversight.
• Intellectual-property infringement or agents built from stolen or unlawfully obtained data.
• Spam, fake reviews, rating manipulation, fabricated testimonials, referral abuse, bot purchases, or fee evasion.
• Marketplace bypass — listing as “free” a product that requires payment to you to use (a paywall, demo-request, or lead-capture destination), or using a listing primarily to route subscribers off-platform to complete a purchase the listing offers. A free listing must be genuinely usable at no charge; a product subscribers must pay you for must be listed as paid. (A genuine free tier with paid upgrades is welcome when the extra costs are disclosed on the listing.)
• Evading a suspension, creating replacement accounts, hiding ownership, or relisting a banned agent.

At launch we prohibit these higher-risk categories outright rather than attempting to review them case-by-case. We may open a documented review path for some of them later, as the marketplace matures and the appropriate safeguards and expert review are in place.

Restricted — manual review required

These are not prohibited outright, but require documented review, appropriate disclosures, minimum controls, and approval before publication. (Where an agent crosses into the prohibited categories above — for example, individualized regulated advice, crisis/safety-critical use, or being directed at children — it is not permitted at launch, not merely restricted.)

• General, clearly-labelled educational or informational content that touches on health, legal, finance, tax, immigration, employment, or safety topics — without giving individualized advice or inviting reliance.
• Agents with autonomous tool use, code execution, external access, email/purchasing, or real-world actions.
• Agents processing sensitive personal information or confidential files.
• Agents making performance, income, compliance, diagnosis, outcome, or savings claims a customer could rely on.
• Any category Stripe identifies as restricted or requiring prior approval.

Third-party AI providers

If your agent relies on another company's AI (OpenAI, Anthropic, Google, and others), you must comply with that provider's terms. Building a product on a commercial API is permitted; reselling access to a consumer account, a shared login, or a personal subscription is not. For any paid agent you must declare one of the following:

• It runs on a paid AI API or business plan that allows selling to customers (like the OpenAI or Anthropic API).
• I run the AI myself, on my own servers or device.
• The AI provider gave me written permission to resell or share it.
• Each customer uses their own AI account. I'm selling my own setup, instructions, or content — not access to my account.

You may not:

• Sharing consumer login credentials (ChatGPT, Claude, Gemini, etc.) across customers.
• Routing multiple paying customers through one personal subscription or plan.
• Selling access to a consumer account.
• Claiming affiliation with OpenAI, Anthropic, Google, or another provider.
• Charging solely for a private shared link where the provider's terms don't permit paid resale.
• Selling paid access to consumer Copilot or consumer Grok — neither has a supported path for paid third-party access (Copilot Studio and the xAI API are the commercial routes).
• Bypassing a provider's usage limits or access controls.

Agents listed under a consumer app are held for confirmation that the provider's terms permit paid resale before they are published. Paid listings on consumer Copilot or consumer Grok aren't accepted at all — neither surface has a supported path for paid third-party access (Copilot Studio and the xAI API are the commercial routes).

What each provider's terms say

Our plain-language reading of each provider's published terms, last reviewed July 5, 2026. This is a summary, not legal advice — the provider's own documents govern, and creators remain responsible for complying with them.

• ChatGPT: OpenAI's terms let you sell a product you build on their paid API. They don't authorize charging for access to a GPT on chatgpt.com — a GPT is contractually part of OpenAI's service, OpenAI can remove any GPT at any time, and sharing a GPT gives OpenAI a license to distribute it to other users. Paid listings that deliver a chatgpt.com link are held for human review.
• Claude: Anthropic's terms let you sell a product you build on the Claude API. They don't authorize charging for access to anything on claude.ai — published artifacts are public by design (no paid or private tier exists) and the consumer terms prohibit reselling the service. Paid listings that deliver a claude.ai link are held for human review.
• Gemini: Google's terms let you sell a product you build on the Gemini API (use the paid tier — the free tier trains on your data). They don't address charging for a shared Gem, and a link-shared Gem opens for anyone with the link without signing in, so a paywall can't actually be enforced. Paid listings that deliver a Gem link are held for human review.
• Copilot: Microsoft retired consumer Copilot GPTs in 2024, and consumer Copilot is personal-use only. The supported commercial path is Copilot Studio: agents you publish there can be used by your customers without a Microsoft license (you pay the usage credits). Paid listings that deliver a consumer-Copilot link aren't accepted — Copilot Studio agents go through normal review.
• Grok: xAI's enterprise terms let you sell a product you build on the Grok API. Consumer Grok has no shareable agent surface — the only way to hand someone your Grok setup is sharing your account, which xAI prohibits. Paid consumer-Grok listings aren't accepted; API-built Grok products (listed under their own platform) go through normal review.
• GitHub: GitHub-hosted Copilot agents weren't part of our July 2026 provider review yet. Paid listings that deliver a github.com link are held for human review while we confirm GitHub's terms.

Enforcement

Every agent starts in review and is not listed or able to take payments until approved. Following a limited eligibility review, we may list, reject, restrict, unlist, suspend, remove, or permanently ban an agent or account, and may disable a high-risk agent immediately while we investigate. Anyone can report an agent or review. A basic appeal path is available. See our Safety page.

Effective date: August 1, 2026.