Generative AI Prohibited Use Policy

Google · Official agreement

Read the agreement

14 of 14 cards

Legal Rights & Agreement Rules · Context

The restrictions apply to your use of generative AI in Google products that refer to this policy

Google says it expects you to use generative AI models in a responsible, legal and safe manner. The restrictions in this policy apply to your interactions with generative AI, but only in the Google products and services that refer to this policy. The page shows a last-modified date of December 17, 2024.

Supporting quotations

Last Modified: December 17, 2024
We expect you to engage with them in a responsible, legal, and safe manner. The following restrictions apply to your interactions with generative AI in the Google products and services that refer to this policy.

Qualifications & exceptions

  • Applies only in Google products and services that refer to this policy.
  • Covers your interactions with generative AI.

Not stated here

  • The policy does not list which Google products and services refer to it.
  • The policy does not say what happens if someone breaks a restriction.

Cross-references

Legal Rights & Agreement Rules · Restriction or exception

You may not generate or distribute content on child sexual abuse, extremism, intimate imagery or self-harm

The policy lists several kinds of content you may not generate or distribute. They are content that relates to child sexual abuse or exploitation, content that facilitates violent extremism or terrorism, content that facilitates non-consensual intimate imagery, and content that facilitates self-harm. The child sexual abuse item says content that "relates to" the subject, which is a wider test than "facilitates" in the other three items.

Supporting quotations

This includes generating or distributing content that:
Relates to child sexual abuse or exploitation.
Facilitates violent extremism or terrorism.
Facilitates non-consensual intimate imagery.
Facilitates self-harm.

Qualifications & exceptions

  • The policy introduces this list with "This includes", under a general ban on dangerous or illegal activities.

Not stated here

  • The policy does not define the terms it uses, such as violent extremism or self-harm.
Also annotated in the same passage (4)

Legal Rights & Agreement Rules · Restriction or exception

You may not engage in dangerous or illegal activity or generate content that facilitates it

The policy tells you not to engage in dangerous or illegal activities or otherwise violate applicable law or regulations. It says this includes generating or distributing content that facilitates illegal activities or violations of law. Its example is providing instructions for synthesizing or accessing illegal or regulated substances, goods, or services.

Supporting quotations

Do not engage in dangerous or illegal activities, or otherwise violate applicable law or regulations.
Facilitates illegal activities or violations of law -- for example, providing instructions for synthesizing or accessing illegal or regulated substances, goods, or services.

Qualifications & exceptions

  • The listed example illustrates the ban and does not limit it ("for example").
  • The ban on content covers content that facilitates illegal activity, and the opening sentence also covers your own activities.

Not stated here

  • The policy does not say which laws or regulations apply or how it decides that content is illegal.
  • The policy does not define "dangerous" beyond the examples it gives.
Also annotated in the same passage (4)

Content & IP Rights · Restriction or exception

You may not generate content that violates others' rights, including privacy and intellectual property

The policy lists content that violates the rights of others, including privacy and intellectual property rights, among the content you may not generate or distribute. Its example is using personal data or biometrics without legally-required consent.

Supporting quotations

This includes generating or distributing content that:
Violates the rights of others, including privacy and intellectual property rights -- for example, using personal data or biometrics without legally-required consent.

Qualifications & exceptions

  • The consent example refers to consent that is legally required, so it does not reach every use of personal data or biometrics.
  • The listed example illustrates the ban and does not limit it ("for example").

Not stated here

  • The saved text shows this item as a separate list entry without nesting. This card reads it as part of the first prohibition's list, because the next prohibition begins at "Do not compromise the security".
Also annotated in the same passage (4)

Use & Processing · Restriction or exception

You may not generate content that makes harmful automated decisions without human supervision in high-risk areas

The policy lists content that makes automated decisions with a material detrimental impact on individual rights, without human supervision, in high-risk domains. It gives employment, healthcare, finance, legal, housing, insurance and social welfare as examples of high-risk domains. The restriction ties to decisions made without human supervision, and this item says nothing about decisions that have human supervision.

Supporting quotations

This includes generating or distributing content that:
Makes automated decisions that have a material detrimental impact on individual rights without human supervision in high-risk domains -- for example, in employment, healthcare, finance, legal, housing, insurance, or social welfare.

Qualifications & exceptions

  • Only decisions with a material detrimental impact on individual rights.
  • Only decisions made without human supervision.
  • Only in high-risk domains; the listed domains are examples ("for example").

Not stated here

  • The policy does not define "material detrimental impact" or give a full list of high-risk domains.
  • The saved text shows this item as a separate list entry without nesting. This card reads it as part of the first prohibition's list.
Also annotated in the same passage (4)

Security & Safeguards · Restriction or exception

You may not generate content that compromises the security of others' or Google's services

The policy tells you not to compromise the security of others' or Google's services. It says this includes generating or distributing content that facilitates spam, phishing or malware, and content that facilitates abuse of, harm to, interference with, or disruption to Google's or others' infrastructure or services.

Supporting quotations

Do not compromise the security of others’ or Google’s services. This includes generating or distributing content that facilitates:
Spam, phishing, or malware.
Abuse of, harm to, interference with, or disruption to Google’s or others’ infrastructure or services.

Qualifications & exceptions

  • The list follows "This includes", so the listed items illustrate the ban.

Not stated here

  • The policy does not define terms such as spam, phishing or interference.
Also annotated in the same passage (1)

Security & Safeguards · Restriction or exception

You may not circumvent abuse protections or safety filters, for example by manipulating the model

The policy lists circumvention of abuse protections or safety filters among the content you may not generate or distribute that compromises security. Its example is manipulating the model to contravene Google's policies.

Supporting quotations

This includes generating or distributing content that facilitates:
Circumvention of abuse protections or safety filters -- for example, manipulating the model to contravene our policies.

Qualifications & exceptions

  • The listed example illustrates the restriction and does not limit it ("for example").

Not stated here

  • The policy does not describe what the abuse protections or safety filters are.
Also annotated in the same passage (1)

Legal Rights & Agreement Rules · Restriction or exception

You may not generate content that facilitates hatred, harassment, bullying or violence

The policy tells you not to engage in sexually explicit, violent, hateful, or harmful activities. It says this includes generating or distributing content that facilitates hatred or hate speech, harassment, bullying, intimidation, abuse or the insulting of others, and violence or the incitement of violence.

Supporting quotations

Do not engage in sexually explicit, violent, hateful, or harmful activities. This includes generating or distributing content that facilitates:
Hatred or hate speech.
Harassment, bullying, intimidation, abuse, or the insulting of others.
Violence or the incitement of violence.

Qualifications & exceptions

  • The list follows "This includes", so the listed items illustrate the ban.

Not stated here

  • The policy does not define hate speech, harassment or violence.
Also annotated in the same passage (1)

Legal Rights & Agreement Rules · Restriction or exception

You may not generate sexually explicit content, such as content made for pornography or sexual gratification

The policy lists sexually explicit content among the content you may not generate or distribute. Its example is content created for the purpose of pornography or sexual gratification.

Supporting quotations

Do not engage in sexually explicit, violent, hateful, or harmful activities. This includes generating or distributing content that facilitates:
Sexually explicit content -- for example, content created for the purpose of pornography or sexual gratification.

Qualifications & exceptions

  • The listed example illustrates the restriction and does not limit it ("for example").

Not stated here

  • The policy does not say whether sexually explicit content for other purposes, such as education or art, falls inside this restriction; a separate sentence allows Google to make exceptions.
Also annotated in the same passage (1)

Legal Rights & Agreement Rules · Restriction or exception

You may not use generative AI for fraud, scams or impersonation meant to deceive

The policy tells you not to engage in misinformation, misrepresentation, or misleading activities. It says this includes frauds, scams, or other deceptive actions. It also includes impersonating an individual, living or dead, without explicit disclosure, in order to deceive.

Supporting quotations

Do not engage in misinformation, misrepresentation, or misleading activities. This includes:
Frauds, scams, or other deceptive actions.
Impersonating an individual (living or dead) without explicit disclosure, in order to deceive.

Qualifications & exceptions

  • The impersonation item applies when you act without explicit disclosure and in order to deceive.
  • The fraud item does not add an intent condition in its wording.

Not stated here

  • The policy does not say what counts as explicit disclosure.
Also annotated in the same passage (2)

Legal Rights & Agreement Rules · Restriction or exception

You may not make misleading claims of expertise, or about government, democracy or health, in order to deceive

The policy lists facilitating misleading claims of expertise or capability in sensitive areas, such as health, finance, government services or the law, in order to deceive. It also lists facilitating misleading claims related to governmental or democratic processes or harmful health practices, in order to deceive.

Supporting quotations

Do not engage in misinformation, misrepresentation, or misleading activities. This includes:
Facilitating misleading claims of expertise or capability in sensitive areas -- for example in health, finance, government services, or the law, in order to deceive.
Facilitating misleading claims related to governmental or democratic processes or harmful health practices, in order to deceive.

Qualifications & exceptions

  • Applies when the claims are made in order to deceive.
  • The sensitive areas named for expertise claims are examples ("for example").

Not stated here

  • The policy does not say how it decides that a claim is misleading or made in order to deceive.
Also annotated in the same passage (2)

Legal Rights & Agreement Rules · Restriction or exception

You may not claim generated content was created solely by a human, in order to deceive

The policy lists misrepresenting the provenance of generated content by claiming a human alone created it, in order to deceive, among the misleading activities you may not engage in.

Supporting quotations

Do not engage in misinformation, misrepresentation, or misleading activities. This includes:
Misrepresenting the provenance of generated content by claiming it was created solely by a human, in order to deceive.

Qualifications & exceptions

  • Applies when the claim is made in order to deceive.

Not stated here

  • The policy does not say whether labeling or disclosing AI use is required in other situations.
Also annotated in the same passage (2)

Legal Rights & Agreement Rules · Restriction or exception

Google may make exceptions for educational, documentary, scientific or artistic reasons or substantial public benefit

Google says it may make exceptions to these policies based on educational, documentary, scientific, or artistic considerations, or where harms are outweighed by substantial benefits to the public.

Supporting quotations

We may make exceptions to these policies based on educational, documentary, scientific, or artistic considerations, or where harms are outweighed by substantial benefits to the public.

Qualifications & exceptions

  • Google may make exceptions; the policy does not promise them.
  • Exceptions rest on educational, documentary, scientific or artistic considerations, or on substantial public benefit outweighing harms.

Not stated here

  • The policy does not say how to ask for an exception or who decides.
  • The policy does not say which of the restrictions an exception can reach.

Cross-references