Thank you for your interest in creating Copilot GPTs using our Copilot GPT Builder. While the Code of Conduct section of the Microsoft Services Agreement governs use of Copilot, we believe creators of Copilot GPTs will benefit from having a more specific explanation of how the Code and MSA apply to this product. The following policies are designed to ensure a safe experience for you and others who will access your custom Copilot GPTs.
These Copilot GPTs Policies apply to:
All information (including your Copilot GPT name and description), instructions, knowledge, documents, content, and data (“Content”) that you provide in connection with the creation of a Copilot GPT;
The intended or reasonably foreseeable purpose of your Copilot GPT(s) and outputs it generates based on your Content.
Misrepresentation and impersonation. Do not build Copilot GPTs that mislead or attempt to mislead or misinform others, including:
Misrepresenting the purpose, ownership, rights, developer information, security and privacy settings, provenance, and any other material information about your Copilot GPT
Attempting to create, promote, or facilitate participation in spamming, scams, phishing, fraud, and other fraudulent activities
Impersonating or conveying a false or misleading affiliation with any other individual, company, entity, brand, or other third parties by using their name, image, likeness, voice, branding, designs, or any other identifying information or content associated with them without their authorization or permission
Attempting to assist in plagiarism or academic dishonesty
If your GPT is intended to provide content related to information, news, or current events in the real world, it must not use or distribute false, misleading, manipulated, or out of date information
Attempting to create fake user profiles, profile data, user accounts, or bots (“Inauthentic Users”) and/or strategies, plans, or instructions intended to create, disseminate, or amplify falsified or fraudulent online engagement (such as likes, reviews, or comments) from Inauthentic Users.
Privacy. Do not build Copilot GPTs that enable, facilitate, or promote activities that could compromise the privacy of others, including:
Collecting, processing, distributing sensitive personal information such as government IDs, banking and payment information, passwords, health data, etc.
Release private, personal or sensitive information of others online without their consent (i.e., “doxxing”)
Jeopardizing or compromising user security or the security or functionality of their device, system, or related systems
Using or attempting to use biometric information for identification or assessment, including facial recognition
Failing to comply with applicable safety and privacy laws, including relating to the collection of personal information from children
Security. Do not create GPTs that facilitate or instruct the creation of malware, viruses, or hacking efforts that could put users, their data, or devices at risk or attempt to undermine Copilot safety controls and restrictions. This includes:
Engaging in bot/scraping behaviors
Engaging in “jailbreaking” or prompt-based manipulation, including efforts to bypass, remove, or reverse engineer Copilot safety restrictions
Sharing, distributing, or attempting to develop spyware, viruses, malware, or other malicious programs to monitor online activities of others or collect personal data or other sensitive information
Attempting to disturb, disable, or disrupt normal operation of Copilot, Copilot GPTs, or any other service of Microsoft and its affiliates and partners
Otherwise jeopardizing the security of Copilot, Copilot GPTs, or their users and devices
Intellectual Property. Do not use Content, APIs, or code, that do not belong to you or that you do not have permission to use to build custom GPTs. This includes:
Using, inputting, uploading, or directing the use of any Content, material or technology you do not have rights to use (including integration of third-party APIs). All Content provided by you for use as part of your custom Copilot GPT (and associated metadata) must be either originally created by you, appropriately licensed from the third-party rights holder, used as permitted by the rights holder, or otherwise permitted by law
Do not build Copilot GPTs that infringe or facilitate, encourage, or promote the infringement of intellectual property rights of others (including copyright, trademark, patent, or publicity rights)
Do not build Copilot GPTs that facilitate, encourage, or promote the sale of infringing or counterfeit goods or services
Do not build Copilot GPTs that facilitate or encourage attempts to bypass technological measures intended to prevent infringement or content sharing
Harmful and Objectionable Activities. Do not build Copilot GPTs that create or distribute content for, facilitate, promote, glamorize, or instruct others to engage in the following harms:
Exploitation and Abuses
Child sexual exploitation and abuse material (CSEAI) (content depicting minor persons engaging in sexually explicit or abusive behavior)
Grooming (the act of an adult communicating with a child for purposes of exploitation)
Non-consensual intimate imagery and intimate extortion, including deepfakes
Sexual solicitation
Human trafficking
Self-Harm
Suicide
Self-injury
Eating disorders
Violence and Coordinated Harm
Attempting or inciting any coordinated harm toward people, animals, or property
Graphic violence and gore
Terrorism and violent extremist content
Violent threats and incitement
Dangerous organizations and individuals
Harmful or Objectionable Content
Hate speech and discrimination, including attacks on individuals or groups based on their protected traits including race, ethnicity, national origin, gender, gender identity, sexual orientation, religious affiliation, age, disability status, caste, or any other characteristic that is consistently associated with systemic prejudice or historical marginalization)
Bullying and Harassment (i.e., content and/or conduct that targets individuals or groups with threats, intimidation, promotion of physical harm, or other abusive behavior)
Activities impacting fundamental rights or social welfare. Do not build Copilot GPTs that could significantly impact user safety, fundamental rights, or wellbeing, including:
Providing tailored or personalized legal, medical, or financial advice or engaging in unlicensed professional activities of any kind
Providing automated decision-making or review in high-stakes areas with potential to significantly affect an individual’s safety, fundamental rights, or well-being, including loan, credit, mortgage, and financial product eligibility, judicial review, law enforcement, employment, housing, education admission, insurance eligibility, military use, or facial recognition
Promoting or encouraging content or activities that significantly contribute to the systemic prejudice, marginalization, or discrimination of individuals or groups
Creating political and/or election-related content that:
Engages in political lobbying or campaigning, including the creation, marketing, or dissemination of campaign materials
Attempts to target, profile, or collect data on specific individuals or demographics in connection with political campaigns or elections
Provides misleading information concerning elections (including misrepresenting voting requirements or processes) or impersonating candidates or nominees for public office
Attempts to deter or discourage individuals from voting or participating in democratic processes
Restricted Goods and Services. Do not build Copilot GPTs offering or promoting restricted or highly regulated goods and services, for example:
Facilitating the sales, transfer, trade, marketing, gifting, or creation of regulated products, including tobacco, alcohol, illegal drugs, hazardous goods or materials, or pharmaceutical products
Facilitating the sales, transfer, trade, marketing, gifting, or creation of firearms, explosives, and other dangerous products intended to cause harm to individuals, animals, or property
Facilitating real-money gambling
Facilitating payday lending
Facilitating the mining or trading of cryptocurrency
Inappropriate Content. Do not build Copilot GPTs intended to create, promote, or facilitate the distribution of inappropriate content, including:
Adult content (including pornography or nudity)
Sexually suggestive or explicit content outside of educational or scientific context
Excessive profanity
Virtual bots or personas intended to foster an online romantic or sexual companionship or relationship with users (i.e., “virtual girl/boyfriends”)
Content that depicts excessive or irresponsible use of drugs, alcohol, or tobacco products
Underage users. Do not build Copilot GPTs targeting or primarily intended for users under the age of eighteen.
Illegal activities. Do not build Copilot GPTs that offer, enable, encourage, facilitate, or promote illegal activities of any kind. Your Copilot GPTs must comply with all applicable laws and regulations.
Advertising and Sponsored Content. Do not build Copilot GPTs:
Where the primary purpose of the Copilot GPT is to get users to click ads, affiliate links, link farms, or spam content
That interfere with or diminish the visibility, value, or quality of any ads that may appear in Copilot GPTs
That fail to adequately disclose the existence of material connections, such as financial incentives, payments, sponsorship, rewards, gifts, complimentary experiences, or business relationships, you have with products or services you promote, discuss, or recommend in your Copilot GPT.
Enforcement
These policies may be enforced at Microsoft’s discretion, and Microsoft may remove, suspend, or disable access to Copilot GPTs at any time without prior notice. Copilot GPTs believed to be in violation of these Policies may be removed or blocked from publication automatically upon submission or retroactively suspended or removed upon further review.
If you encounter a Copilot GPT or content generated from a Copilot GPT that violates these policies, please submit a report using the Report a Concern.
