Petition to Companies That Develop and Operate Generative AI Image Systems
The human body is not inherently harmful, obscene, or dangerous. Yet many generative artificial intelligence systems continue to treat the visible human body as though nudity itself were evidence of sexual content, exploitation, or abuse.

Artists, educators, historians, health advocates, body-acceptance organizations, and ordinary users regularly encounter refusals when they attempt to create, edit, restore, analyze, or discuss images involving nonsexual nudity. A request may concern fine art, anatomy, swimming, bathing, historical life, body diversity, disability, aging, health education, documentary work, or an ordinary unclothed person in a nonsexual everyday setting. Nevertheless, the system may block the request simply because parts of the human body are visible.

This is not careful safety moderation. It is a failure to make necessary distinctions.

We call on companies that develop and operate generative AI image systems—including OpenAI, Google, Microsoft, Adobe, and other major providers—to create and consistently implement policies that distinguish lawful, nonsexual, and non-exploitative depictions of human nudity from sexual conduct, sexualization, exploitation, harassment, coercion, and abuse.

Anatomy Is Not an Action
A body is not a behavior.

Nudity may appear in sexual content, but neither human anatomy nor nudity is synonymous with sexual activity. The presence of breasts, buttocks, or genitals does not by itself establish that an image is erotic, exploitative, pornographic, coercive, or harmful.

Context matters.

A nude figure in an art class is not the same as a sexualized image. A person swimming without clothing is not the same as a person engaged in sexual conduct. A medical illustration is not pornography. A historical photograph is not exploitation merely because the people in it are unclothed. An ordinary human body standing, resting, bathing, changing, receiving care, or participating in a nonsexual social setting should not automatically be classified as dangerous content.

AI systems are already designed to evaluate complex combinations of language, imagery, identity, behavior, and context. These systems should be developed to make better distinctions rather than relying on the mere visibility of anatomy as a substitute for thoughtful classification.

The Missing Middle
Current restrictions often erase the large space between fully clothed commercial imagery and explicitly sexual content.

That missing middle includes ordinary human bodies across the lifespan and across the full diversity of size, ability, race, appearance, physical development, scarring, aging, and health. It includes bodies shown in art, education, recreation, medicine, documentary work, historical settings, family life, and everyday experience.

When AI systems repeatedly treat ordinary anatomy as unsafe, they reinforce the cultural assumption that the unclothed body is automatically sexual, shameful, threatening, or inappropriate. They also limit the ability of responsible creators to represent real human life with honesty and dignity.

This affects more than artists. It affects educators trying to explain anatomy without panic, advocates working to reduce body shame, historians examining cultures with different clothing norms, health communicators representing diverse bodies, and ordinary people trying to understand their relationship with the human form.

The result is not genuine neutrality. It is a technological preference for concealment.

Safety and Body Acceptance Are Compatible
This petition does not call for unrestricted image generation.

Strong protections remain necessary against genuine abuse. AI companies should continue to prohibit sexual conduct involving minors, all sexualized or exploitative depictions of minors, non-consensual intimate imagery, voyeuristic content, coercion, sexual exploitation, targeted humiliation, identity-based sexual deepfakes, harassment, and other forms of abuse.

Sexualization does not require the depiction of a sexual act. It can also occur through provocative or suggestive posing, voyeuristic presentation, fetishistic or degrading framing, unnecessary emphasis on intimate anatomy, identity misuse, or imagery created for sexual gratification.

These protections are essential and must remain fully enforceable.

But the need to prevent exploitation does not justify treating every unclothed human body as though exploitation has already occurred. Effective safety systems should distinguish between a body and an abusive use, sexualization, or exploitation of that body.

Child safety must be based on what is being done, how a person is being represented, whether the context is sexualized or exploitative, whether privacy and identity are being protected, and whether coercion, abuse, or malicious intent is involved. The mere presence of ordinary human anatomy is not itself sexualization.

Overbroad restrictions can burden legitimate users while failing to address the factors that actually determine whether content is harmful: what conduct is occurring, the age and capacity of those depicted in relation to conduct and consent, whether valid or appropriate consent or authorization exists, whether anyone is being coerced or exploited, whether a real identity is being misused, and whether the framing is sexualized, voyeuristic, degrading, or malicious.

The answer is not weaker safety. It is more accurate safety.

What We Are Asking AI Companies to Do
We ask companies that develop generative AI image systems to adopt the following reforms.

1. Establish a Clear Category for Nonsexual Human Nudity
Published policies should explicitly recognize lawful, nonsexual, and non-exploitative depictions of human nudity as distinct from sexually explicit, sexualized, abusive, or exploitative content.

This category should include appropriate artistic, educational, historical, documentary, medical, health-related, body-acceptance, recreational, family-life, and ordinary-life contexts.

Policies should not rely on exceptions that are recognized in principle but applied inconsistently by the product itself. The distinction must be implemented consistently in image generation, image editing, image analysis, restoration, and other relevant tools.

2. Evaluate Context, Conduct, Consent, Age, and Representation
Moderation systems should evaluate what is happening in an image and how people are being represented rather than treating visible anatomy as conclusive evidence of harm.

Relevant factors should include whether the context is sexual or nonsexual, whether sexual conduct is present, the age and capacity of those depicted in relation to conduct and consent, whether any person is being sexualized or exploited, whether appropriate consent, authorization, privacy, and identity protections exist, whether a real person’s identity is being misused, whether the framing is provocative, voyeuristic, fetishistic, or degrading, and whether the request involves harassment, coercion, or abuse.

The involvement of children must trigger especially stringent protections against sexualization, exploitation, coercion, identity misuse, voyeurism, and abuse. Those protections should evaluate conduct and representation, not classify ordinary anatomy itself as sexual content.

Anatomy should be one contextual feature, not the automatic verdict.

3. Permit Responsible Artistic, Educational, Historical, Medical, and Documentary Use
Users should be able to responsibly generate, edit, analyze, restore, and transform lawful material involving nonsexual and non-exploitative human nudity.

This should include fine art, art education, medical and anatomical education, historical representation, documentary work, body-diverse imagery, disability representation, aging, pregnancy, bathing, swimming, caregiving, and other ordinary nonsexual situations in which the unclothed body is relevant.

Copyright, appropriate consent or authorization, privacy, identity protection, and strict safeguards against the sexualization, exploitation, abuse, and identity misuse of minors should remain fully enforceable.

4. Explain Refusals Clearly
When a request is blocked, the system should provide a meaningful explanation of the policy category involved.

A user should be able to tell whether a request was rejected because of suspected sexual content, possible sexualization of a minor, identity misuse, non-consensual imagery, exploitation, voyeurism, or a general product restriction.

Generic warnings that merely label a request “unsafe” conceal the distinction between harmful conduct and ordinary anatomy. They also make it difficult for responsible users to revise a request or report a false positive.

5. Distinguish Good-Faith False Positives from User Misconduct
A blocked request should not automatically be treated as evidence that the user attempted to violate safety rules.

People making legitimate requests involving art, health, history, education, recreation, documentary work, or body acceptance should not be made to fear account penalties because an automated classifier misunderstood their request.

Companies should clearly distinguish a routine refusal from a confirmed policy violation, especially when the content is ambiguous or context-dependent.

Good-faith attempts to depict or discuss ordinary human anatomy should not be treated as equivalent to attempts to generate sexualized, coercive, exploitative, or abusive content.

6. Provide Meaningful Feedback and Appeal Mechanisms
Users should have a practical way to report recurring or obvious classification errors.

Not every refusal can receive individual human review, but companies can create structured systems for collecting disputed examples, identifying patterns, testing model behavior, and improving policies and classifiers.

Companies should also publish aggregate information about common false positives, policy changes, appeal outcomes, and improvements in contextual moderation.

7. Test for Unequal and Inconsistent Treatment
AI companies should test whether their systems treat people differently based on sex, gender presentation, body size, disability, age, race, scarring, pregnancy, physical development, or other bodily characteristics.

Moderation rules should not reproduce familiar cultural inequalities—for example, treating equivalent male and female anatomy differently without a clearly stated and defensible reason.

Safety testing should include ordinary, non-idealized bodies and a wide range of legitimate nonsexual contexts rather than relying primarily on polished commercial imagery.

Testing should also examine whether systems confuse age with sexualization or treat the mere visibility of a younger person’s anatomy as proof of sexual content while failing to evaluate posing, framing, conduct, identity use, and intent accurately.

Why This Matters
Every human life is lived through a body.

Technology increasingly influences which bodies can be represented, which experiences can be discussed, and which images are considered acceptable. When the systems used to create modern culture repeatedly conceal ordinary anatomy, they do more than decline an image request. They help establish the boundaries of what society is permitted to see.

Those boundaries should not be governed by the assumption that an unclothed human body is inherently sexual or unsafe.

The body can be vulnerable without being shameful. It can be private without being obscene. It can appear in sexual contexts without being sexual in every context. It can be treated with boundaries, consent, care, and dignity without being erased.

Children must be protected absolutely from sexual conduct, sexualization, exploitation, coercion, voyeuristic or provocative depiction, identity misuse, and abuse. But protection should not require teaching that ordinary human anatomy is itself the source of danger.

A person’s body is not the violation. Exploitation, coercion, sexualization, and abuse are the violations.

Artificial intelligence should be capable of recognizing those distinctions.

Why Your Signature Matters
A petition alone cannot rewrite an AI company’s policies or guarantee a change in product behavior. It can, however, document public demand, give affected users a shared statement, support direct outreach to technology companies, encourage journalists and researchers to examine the problem, and demonstrate that overbroad body censorship is not a minor inconvenience affecting only a handful of people.

It can also encourage a more serious public examination of whether safety policies are being judged by their actual ability to prevent exploitation or simply by how completely they suppress visible anatomy.

Each signature strengthens the case that users want safety systems based on context, conduct, consent, representation, and actual harm—not the automatic classification of human anatomy as dangerous.

By signing, you are asking AI companies to develop more accurate, transparent, and humane moderation systems.

You are not asking them to abandon safety.

You are asking them to distinguish safety from shame.

Sign the Petition
We call on OpenAI, Google, Microsoft, Adobe, and other developers of generative AI image systems to establish and consistently implement a distinct policy category for lawful, nonsexual, and non-exploitative human nudity.

We ask these companies to evaluate conduct, context, consent, age, capacity, identity use, framing, and intent rather than body parts alone. We ask them to protect legitimate artistic, educational, historical, medical, documentary, recreational, and body-acceptance uses; explain refusals clearly; improve feedback and appeal systems; and maintain strict safeguards against sexual conduct involving minors, sexualization, exploitation, coercion, abuse, non-consensual imagery, voyeurism, and identity misuse.

The human body is a baseline of reality, not a safety hazard.

Anatomy is not an action.

This petition is an initiative of ShameProof.life, a project created to encourage thoughtful examination of how culture shapes our feelings about the human body and to help people begin more honest conversations about body comfort, dignity, privacy, safety, and acceptance.

After signing, visit ShameProof.life to explore the questions behind this campaign and help start a wider conversation about how we see, represent, protect, and treat the human body.