Community Guidelines

Effective date: 11 June 2026

1. Welcome

mu is a European social application built on the AT Protocol. It is operated by Stichting Modal, a non-profit Dutch foundation, as part of the Eurosky programme.

These Community Guidelines describe how we expect people to use mu, how we moderate content, and what you can expect from us when something goes wrong. They are part of your agreement with us under the mu Terms of Service; words in capitals that we don't define here are defined there.

mu is for adults. The minimum age is 18.

2. Our values

mu takes positions on certain issues. We are transparent about those positions. The guidelines that follow flow from four values. They are the guiding principles we follow when we make difficult moderation calls.

Plurality. A healthy public sphere has many spaces with many norms. mu is one space among many on the open social web. We do not expect to be the right place for everyone; we try to be a good place for our users, and we make it easy for people to take their data and connections elsewhere.

Healthy discourse. Disagreement, criticism, and discomfort are part of public conversation, and we protect them. Harassment, threats, dehumanisation, and coordinated abuse don't have a place in healthy public conversation, and we act against them. The line between robust debate and abuse is the line we work hardest to get right.

Public interest. mu has a public interest mission. We make decisions that serve the values of democracy and human rights, even when that costs us short-term growth or engagement.

Openness. Your data is yours; your followers are yours; your identity is yours. You can leave mu and take everything with you. Our moderation is transparent, our decisions are explained, our processes are appealable.

3. How moderation works on mu

Moderation on mu has two layers, plus the tools you have yourself.

The Bluesky baseline. We ingest the moderation labels published by Bluesky Social and apply them on mu by default. That is a practical choice for launch. Operating a full trust-and-safety classification stack at scale costs money and people we do not yet have, and Bluesky's labels are a reasonable starting point for the open social web. Bluesky is named here for honesty about the baseline; we do not have a contract with them, and they take no responsibility for moderation on mu.

mu's own layer. On top of the baseline, we act independently. Under these guidelines, we can label, restrict, hide, or remove content and accounts on mu, decline to apply or actively reverse a Bluesky label where we conclude it does not serve our users or does not reflect European norms, and apply additional restrictions in specific Member States where local law requires.

Our direction of travel. mu plans to move to fully independent moderation in two stages.

By summer 2026: mu's own moderation layer launches on top of the Bluesky baseline. This lets us apply specific actions (label, restrict, hide, remove) at the mu layer, decline or reverse Bluesky labels, and apply Member State restrictions.
Within twelve months of launch: mu's own appview launches and we step away from the Bluesky baseline. From that point, mu operates without depending on Bluesky's moderation signals or content delivery.

The baseline relationship with Bluesky exists because we are launching small, not because we want to be downstream of someone else's moderation decisions.

What we can do at launch. Until our own moderation layer is live, what we can do at the mu layer is more limited than the rules above describe. We can suspend an account from signing in to mu where suspension is warranted. We apply Bluesky's labels as the systematic moderation signal on mu. We comply with binding legal orders. We cannot yet take specific moderation actions (hide, restrict, label, remove) against specific posts at the mu layer; the most we can do at the network level is flag the case back to Bluesky's Trust & Safety team, with whom we are in regular contact. The rules in §§4–7 describe the standard we hold ourselves to and the actions we will take once we are able.

Suspension and content visibility. At launch, suspending an account from signing in to mu does not, by itself, remove that account's content from mu. mu uses Bluesky's appview to deliver content, so public content from any account on the network (including suspended accounts and accounts that have never used mu) may continue to appear on mu unless we take a separate moderation action against the specific content. Once our own moderation layer is live (summer 2026), we can apply specific actions to specific posts on mu. Once our own appview is live (within twelve months of launch), suspension means the suspended account's content stops appearing on mu, because we control what our appview serves.

In some EU Member States, additional restrictions apply by law (for example, around hate speech, genocide denial, or specific terrorist designations). Where local law requires, we apply additional restrictions to users in that Member State.

The mu Terms of Service set out the layered moderation architecture in legal detail. See §23 of the Terms.

4. Safety: protecting people from severe harm

These rules describe the categories where the effect of getting it wrong is highest. For these categories we move as quickly as we can and apply the strongest enforcement. We address harassment, which is generally the category with the largest volume of issues, in the next section.

4.1 Child safety

mu does not host accounts of people under 18. We take prevention of child sexual exploitation very seriously.

You may not create, share, promote, or solicit:

child sexual abuse material (CSAM), in any form, including computer-generated, illustrated, or animated versions;
sexualised content that depicts or appears to depict minors, even where it does not amount to legally classified CSAM;
content that promotes contact between adults and minors for sexual purposes, in any form.

You may not use mu to attempt to contact, identify, or share personal information about a minor for any unlawful purpose.

We report CSAM to relevant authorities under EU and Member State law, and we cooperate with the relevant reporting bodies.

4.2 Terrorism and violent extremism

You may not use mu to coordinate, promote, recruit for, or distribute content on behalf of terrorist organisations or violent extremist movements.

This category includes the categories of terrorist content covered by Regulation (EU) 2021/784. Where we receive a removal order under that Regulation we will act within one hour, as the Regulation requires.

You may not share content that glorifies terrorism, recruits for terrorist activity, or provides operational instruction for terrorist acts. Documentary reporting, counter-extremism research, and historical or academic discussion of these subjects with clear context are allowed; see §9 protected expression.

4.3 Threats, incitement, and graphic violence

You may not threaten others with violence or death, incite violence against an individual or group, or post graphic content of real-world violence in order to shock, intimidate, or celebrate harm. For the avoidance of doubt, making jokes about any of these can also constitute a violation of the community guidelines, context dependent; see Appendix.

Combat sports, regulated extreme sports, news and documentary footage, and discussion or depiction of violence in clearly fictional or artistic contexts are allowed.

4.4 Self-harm and mental health

The rule we enforce is simple: we remove content that promotes, instructs in, or glorifies self-harm, suicide, or disordered eating. We do not act against people sharing recovery experiences, support, or survivor stories. See §9.

If we believe someone on mu is at immediate risk to themselves, we may take action and provide regionally appropriate crisis resources.

4.5 Animals

You may not post content depicting cruelty to animals, organised animal fighting, or sexual content involving animals, including realistic computer-generated versions. Lawful activities such as hunting, fishing, and farming are not covered by this rule.

5. Respect for other people, their rights, and the law

This section covers the rules that govern how you treat other people on mu and the lines drawn by European law. We do not separate "social norms" from "legal requirements" here, because in practice they overlap and the most consequential rules sit at the intersection. We do not draw the line at whether such conduct is unlawful or not; we also focus on the question of whether it creates a hostile environment for the mu community.

5.1 Hate speech and denial of historical atrocities

mu does not remove content simply because it is offensive or socially objectionable. We do remove content that crosses into incitement to violence, threats, or dehumanisation, and we remove categories of expression that European democracies have long agreed are not legitimate public discourse. In many cases such speech is unlawful and falls outside the right to freedom of expression.

You may not use mu to:

attack or dehumanise people on the basis of their race, ethnicity, national origin, religion, caste, sex, gender identity, sexual orientation, disability, refugee or migration status, or another protected characteristic;
incite violence or discrimination against any of the above groups;
engage in coordinated harassment campaigns targeting members of the above groups;
deny, distort, or grossly minimise the Holocaust or other historically recognised genocides.

Note: there is no definitive list of genocides. We look at pronouncements by UN and European bodies (for example the General Assembly and the European Court of Human Rights) as well as the context of the discussion. Having a genuine discussion about whether atrocities reach the legal threshold of a genocide, whilst still acknowledging they are atrocities, is acceptable. However, denying that the Holocaust or the Srebrenica genocide (this is a non-exhaustive list) have occurred is never okay: they have been widely recognised as genocides, including by the UN.

Speech that is lawful but offensive is not removed under this section. We may apply labels or reduce surfacing in our recommender feeds. Third-party label services may further filter it (see §12). In short: if it is lawful, we do not remove; we may label or downrank it.

5.2 Harassment and coordinated abuse

You may not target an individual or a group on mu with persistent unwanted contact, threats, slurs, intimidation, sexualised abuse, or coordinated pile-on behaviour. You may not use mu's features (lists, replies, quote posts, blocks, labels) to facilitate or amplify abuse.

We treat coordinated harassment that targets a person or group on the basis of a protected characteristic as a more serious form of this rule, overlapping with §5.1. mu is open to everyone, and as such behaviour that aims to exclude a specific group based on race, gender, sexual orientation, and other protected characteristics, is taken very seriously.

You may strongly disagree with people in public. Criticism, refutation, satire, ridicule of arguments and of public figures' conduct in public life are allowed. The test is whether the activity is aimed at silencing, terrorising, or humiliating an individual rather than at engaging with their ideas or public conduct.

These rules don't just apply to posts; they apply to other content such as the language on profiles and block lists. If that language itself amounts to harassment, it may be blocked.

5.3 Privacy: personal data, doxing, identification

mu sits in a European data-protection culture and we apply a strict line on privacy.

You may not publish or share another person's:

home address, workplace address, or precise current location;
private telephone number, private email address, or other contact details not made public by that person;
identity documents, government identifiers, or other personal data the person has not chosen to make public;
private communications, intimate imagery, or other materials that the person has a reasonable expectation will remain private;
identification of a pseudonymous user, journalist, activist, or whistleblower, where the disclosure would cause foreseeable harm.

Public-interest reporting (for example, identifying a public official engaged in misconduct, in a way and to a degree proportionate to the public interest) is not covered by this rule. See §9.

5.4 Non-consensual intimate imagery and synthetic intimate imagery

This category is treated as a standalone priority because EU law treats it that way, and because the harms are severe and time-sensitive.

You may not create, share, threaten to share, or promote tools to create intimate imagery (including nude, sexual, or sexually suggestive imagery) of any person without that person's express, informed consent. This applies whether the imagery is real, edited, computer-generated, or otherwise synthetic.

We act on credible reports under this rule on a priority basis, including on reports from people who do not have a mu account (see §10 below).

5.5 Regulated goods and services

You may not use mu to sell, offer for sale, advertise, or facilitate transactions in goods or services where doing so would be unlawful in the jurisdiction of the user or of the buyer. This includes, without limitation:

controlled substances and illegal drugs;
prescription medication outside the prescription system of the relevant country;
firearms, ammunition, explosives, and weapon conversion components;
stolen goods, counterfeit goods, and forged documents;
exploitation of protected animal species.

Lawful commercial discussion (for example, vendors of legal products promoting their own legal products) is allowed.

5.6 Intellectual property

You may not post material on mu that you do not have the right to share. That includes copyright-infringing material, trademark violations, and unauthorised redistribution of protected content.

Our Copyright Policy describes the notice format and counter-notice procedures.

5.7 Information integrity

We do not remove "misleading content" generally, however we do act on specific high-harm categories:

coordinated inauthentic behaviour designed to manipulate public discourse, including networks of accounts presenting themselves as independent grassroots voices;
false information about voting procedures, voter eligibility, election dates, locations, or results, where the intent or foreseeable effect is to suppress or distort participation in democratic processes;
false health information that creates a foreseeable risk of immediate physical harm (for example, instructions to ingest poisonous substances framed as remedies);
synthetic media used to deceive about a real person's words or actions, where the deception risks real-world harm. See §6.3.

Our threshold is "demonstrable falsehood with foreseeable serious harm", not "things we think are wrong."

6. Authenticity

Open social networks only work if people can broadly trust that accounts are who they claim to be and that engagement signals are real.

6.1 Impersonation and account authenticity

You may not impersonate another person, an organisation, or an official body in a way that could mislead other users. You may use clearly-labelled parody, fan, or commentary accounts; the test is whether a reasonable user would understand the account is not the person or body it appears to depict.

You may not switch your account into a different identity to keep the previous identity's followers without telling them, register handles you don't plan to use, or undertake coordinated impersonation campaigns.

6.2 Spam and manipulation

You may not use mu to send unsolicited bulk messages, repeatedly post the same content, run engagement farms, manipulate engagement metrics, or operate networks of accounts that pretend to be independent users.

You may not use undisclosed sponsorship or undisclosed commercial content. If you are being paid to post about something, say so clearly.

6.3 Synthetic and AI-generated media

mu does not have a general restriction on your use of AI tools. However, you may not use synthetic media that could be mistaken for real footage or audio without being labelled as such.

You must label any image, audio, or video you post on mu that has been generated or substantially altered using AI or any other tools, where the result depicts a person, real or imagined, in a way that could plausibly be mistaken for real footage or audio of a real person.

We recognise this is hard to verify perfectly, and our enforcement will be best-effort and reactive rather than universal. We will act on reports under this rule and on signals from third-party detection systems where reliable.

We will always act on synthetic media used to deceive about a real person's words or actions in ways that risk real-world harm (election interference, fraud, harassment), under §5.4 (intimate imagery), §6.1 (impersonation), and §5.7 (information integrity), as applicable.

7. Platform and protocol integrity

7.1 Platform and protocol security

You may not attempt to compromise, exploit, bypass, abuse, or disrupt mu's systems, security features, APIs, rate limits, or infrastructure. You may not abuse the AT Protocol itself in ways that degrade the network for other users or operators.

Coordinated disclosure of security vulnerabilities through responsible channels is welcomed and not covered by this rule. See contact details in §15.

7.2 Evasion of enforcement

You may not create new accounts to evade a suspension, ban, or other enforcement action against you. You may not help others evade enforcement.

8. Adult sexual content

mu is for adults and we allow consensual adult sexual content under specific conditions:

all depicted people must be adults and must have consented to the creation and distribution of the content;
the content must be labelled;
the content must comply with §5.4 (no NCII or non-consensual synthetic intimate imagery) and §4.5 (no sexual content involving animals);
the content may not depict realistic risk of death or serious injury, regardless of consent;
§4.1 (Child Safety) always takes precedence.

Creators who post adult sexual content on mu must comply with applicable age-verification and creator-disclosure laws under EU and Member State law.

We act swiftly on credible reports that any of the above conditions has been breached.

9. Protected expression: what we will not moderate

9.1 Absolute prohibitions: no carve-outs

Some categories of content are never permitted on mu, regardless of who is posting and for what purpose. The protections in §9.2 below do not override these. The categories are:

child sexual abuse material in any form, including computer-generated, illustrated, or animated versions (§4.1);
non-consensual intimate imagery of any person, real or synthetic (§5.4);
terrorist content covered by an EU removal order under Regulation (EU) 2021/784 (§4.2);
any content covered by a binding takedown order from a competent court or authority.

Reporting on these categories with appropriate context, and discussing them as a matter of public interest, is allowed. Hosting or republishing the content itself is not, even in a journalistic, research, or educational context.

9.2 Protected categories

Outside the absolute prohibitions in §9.1, these guidelines protect categories of expression that serve the public interest. The behaviours in §§4 to 7 above are not prohibited when the behaviour is part of:

Journalism, research, education. Factual reporting, academic research, and educational content about prohibited subjects, with clear context. A news photo of armed conflict is not a violation of §4.3. An academic paper analysing extremist rhetoric is not a violation of §4.2. A documentary clip showing an act of violence is not a violation of §4.3.

Recovery and survivor accounts. People sharing their own experience of self-harm, eating disorders, addiction, abuse, or violence for the purpose of recovery, support, or advocacy are protected, even where the content describes harms that would otherwise be prohibited.

Public-interest transparency. Information drawn from court records, business filings, regulatory disclosures, and the public conduct of public officials is protected, even where it identifies named individuals. The test is proportionality and public interest.

Artistic, fictional, satirical contexts. Violence, sex, drug use, and other prohibited categories depicted in clearly artistic, fictional, or satirical contexts are protected, with adult content labelled per §8.

Criticism and parody of public figures and institutions. Within the law of the relevant Member State. Strong criticism, unflattering parody, and adversarial questioning of those who exercise public power are protected.

Political speech. Within the law. Especially during election periods, we are conservative about restricting political speech.

When you appeal a decision under §10, telling us which of these protections you believe applies is the fastest way to a useful review.

Protection under §9.2 is protection from mu's moderation. It does not override your responsibility under applicable law, including defamation law. Public-interest transparency, criticism of public figures, and the other categories above shield you from mu acting against your content under these guidelines; they do not shield you from legal claims by named individuals under the law of the relevant jurisdiction. If a court of competent jurisdiction or other competent authority orders us to remove content covered by §9.2, we will comply (see §9.1 absolute prohibitions, which include binding takedown orders).

10. How we enforce

10.1 Statement of reasons

When we take action against your content or your account, we will tell you what we did, why we did it, the specific guideline we relied on, whether the decision was made by a human or by an automated system, and the routes you have to challenge it. We do this in the format required by Article 17 of the Digital Services Act where it applies.

10.2 The enforcement ladder

Our usual ladder, for behaviour that does not require an immediate severe response, runs as follows:

Warning. First-time or low-severity behaviour: a notice from us explaining what crossed the line and what to do differently.
Label or restriction. Labels applied to the content; possible reduction in surfacing in our recommender feeds.
Content removal. Removal of the specific content, with a strike against the account.
Temporary suspension. A defined-length pause on posting and interacting, for repeat behaviour or for individual high-severity incidents that do not require permanent removal.
Permanent suspension. Final removal of the account from mu.

Strikes age out over time; the timeline is set in our internal enforcement procedure.

At launch, steps 2 and 3 (label/restriction at the mu layer and specific content removal) are not yet available; we operate on warnings, suspensions, and the systematic Bluesky baseline labels described in §3. The full ladder becomes available when our own moderation layer is live (target summer 2026).

10.3 Immediate severe action

For some behaviour we skip the ladder and go straight to temporary or permanent suspension without prior warning. That includes:

child safety violations (§4.1);
terrorism and violent extremism content covered by an EU removal order (§4.2);
credible threats against the life or physical safety of an identified person (§4.3);
non-consensual intimate imagery (§5.4);
coordinated inauthentic behaviour designed to manipulate public discourse (§5.7);
significant breaches of platform or protocol security (§7.1);
enforcement evasion under §7.2.

10.4 Appeals

You can appeal any moderation decision against you. Appeals are always reviewed by a person. The internal appeal is described in §26.2 of the mu Terms of Service.

If you are in the EU or EEA you can also take a dispute to a body certified under Article 21 of the Digital Services Act. You can find certified bodies through your country's national Digital Services Coordinator or the European Commission's public list. We will engage in good faith with any certified body that accepts the case.

Nothing in these guidelines affects your right to take a dispute to court.

11. Tools you have

Most disagreements on mu are resolved by the people involved, not by us. Tools we provide:

Block. Stops another account from interacting with you and removes their content from your view. Public.

Mute. Removes another account's content from your view without notifying them or signalling publicly.

Mute words and phrases. Hides posts containing patterns you define, including hashtags and partial phrases.

Choose your feeds. At launch, the default feeds in mu use Bluesky's appview and recommender, surfaced through mu (see §3 and §24 of the mu Terms of Service). You can select alternative feeds, including third-party feeds, or opt out of personalised recommendations altogether. When mu's own appview is live, the default feeds will use our own recommender.

Sensitive-content filters. Choose whether to see content with sensitive labels (adult content, graphic violence, and similar) or to hide it. In addition to our own filters, various third-party labelling and filtering services are available on the AT Protocol that you can subscribe to (see §12 on composable moderation).

Account portability. mu can be used in two ways. If you sign up to mu without an existing Atmosphere account, we create an Eurosky-hosted account for you on the Eurosky Personal Data Server. You can move that Eurosky account to another PDS provider on the AT Protocol at any time, or use it in other ATProto applications. If you already have an account on another PDS (for example, Bluesky's), you can sign in to mu with that account; your account stays where it lives, and you can move on whenever you want. See §20 and §28 of the mu Terms of Service for the mechanics.

12. Composable moderation: third-party label services

mu supports user-subscribed label services run by independent third parties on the Atmosphere. You can subscribe to additional label services that reflect the standards of communities you trust (for example, language-community moderators, topic-specific safety groups, or journalistic verification projects).

Third-party label services are not run by us. We do not endorse them; we do not vouch for the accuracy or fairness of their classifications. They sit on top of mu's own moderation and the Bluesky baseline, never below: a third-party service can add restrictions for you, it cannot remove restrictions that mu has applied.

13. Trusted flaggers

Under Article 22 of the Digital Services Act, entities awarded trusted flagger status by a national Digital Services Coordinator receive priority processing of their notices. mu honours this status.

Trusted flaggers are not above these guidelines; their notices are reviewed under the same rules. They are processed faster and with the assumption of good faith and competence.

The current list of trusted flaggers we recognise, with their certification basis and the categories of content each is certified for, is published in our Transparency Centre. We derive that list from the EU Commission's master list of entities awarded trusted flagger status by national Digital Services Coordinators.

14. Updates to these guidelines

We update these guidelines to reflect changes in the law, in our service, and in our understanding of how to do moderation well. Material changes are announced through mu and by email at least 30 days before they take effect. The current version and its effective date are always shown at the top of this page.

Older versions of these guidelines, and a log of substantive changes, are kept in our Transparency Centre.

15. Contact

To report content or behaviour: in the mu app, use the report button on the post or profile, or email safety@mu.social. The email channel is open to people who do not have a mu account.

To appeal a moderation decision against you: appeals@mu.social.

For copyright and trademark notices: ip@mu.social, in the format described in our Copyright Policy.

For Digital Services Act enquiries from members of the public, including DSA notices of illegal content submitted by recipients of the service: dsa-support@eurosky.tech.

For DSA correspondence primarily for use by Member State authorities, the European Commission, the European Board for Digital Services, and trusted flaggers: notices@eurosky.tech.

For coordinated security disclosure: security@mu.social.

Postal: Stichting Modal, Kranenburgweg 135 A, 2583 ER The Hague, Netherlands.

Appendix: worked examples

This appendix gives concrete examples for each category to help users understand the lines and to give moderators consistent reference points. Examples are illustrative, not exhaustive. In most cases here we remove a post or apply a label as a consequence. In many cases removal will be accompanied by a temporary suspension and, depending on the specific circumstances, a permanent ban.

At launch, our ability to apply these examples on mu is the at-launch subset described in §3 and §10.2: we can suspend accounts but cannot yet apply specific actions to specific posts. The examples describe the standard we will apply once our own moderation layer is live (target summer 2026), and the standard our enforcement against specific accounts already follows.

Child safety (§4.1)

Removed: photo or video that sexualises a minor, regardless of source or context; an account that solicits images from people it believes to be minors; an AI-generated image depicting a minor in a sexualised pose.
Not removed: a parent's photo of their child at a non-sexual public event; a child welfare organisation's awareness campaign material; a news report about a child safety case, with appropriate context.

Hate speech and denial (§5.1)

Removed: a post calling for the deportation of all members of a named ethnic group; a meme dehumanising migrants as animals; a thread denying the Holocaust occurred; a post denying the nationhood of Palestinians; posts calling members of a religious community by slurs.
Not removed: news reporting on hate speech that quotes the speech with context; academic analysis of extremist rhetoric; satire of bigoted positions.

Privacy and doxing (§5.3)

Removed: posting another user's home address without their consent; sharing the legal name of a pseudonymous journalist who has been the subject of state harassment; publishing leaked passport scans of a private individual.
Not removed: publishing the address of a registered office of a corporation as part of legitimate reporting or peaceful campaigning; sharing publicly available information from court records.

Non-consensual intimate imagery (§5.4)

Removed: sexual image of an identifiable person posted without consent, whether original or edited; AI-generated nude image of an identifiable real person; offer to create such imagery for payment.
Removed even without identification: image purporting to be of a named person, regardless of authenticity, where the apparent purpose is to harass or sexualise the named person.

Harassment (§5.2)

Removed: a coordinated reply pile-on directing slurs at an individual or group of people; persistent unwanted sexual messages after the recipient has asked them to stop; threats to contact an individual's employer to harm their livelihood for a non-public-interest reason.
Not removed: a single critical reply, even if rude, to a public figure's public statement; quote-posting a public figure's argument to refute it; collective ridicule of an institution's public position.

Synthetic media (§6.3)

Required to label: an AI-generated portrait that depicts a fictional person realistically; an AI voice cover of a song using a synthetic voice; a generated video in the style of news footage.
Removed: an unlabelled deepfake of a real politician appearing to say something they did not say, intended to be taken as real; a synthetic image used in an impersonation account.
Not removed: clearly captioned AI-generated illustrations, images or videos used for commentary or satire where the output is not mistakable as real.

Information integrity (§5.7)

Removed: a coordinated network of newly created accounts posting claims about a polling station closure; a post falsely claiming a specific medication will treat a serious illness in a way that would foreseeably lead users to harm themselves; a deepfake of an election official appearing to concede.
Not removed: an individual user's mistaken claim about a contested fact; a strong political opinion that we (or others) consider misleading; a public figure's controversial statement reported on for what it is.

Adult content (§8)

Removed: content depicting people who are or appear to be under 18; content posted by or depicting people whose consent has not been given; content depicting extreme violence in a sexual context.
Required to label: any content depicting nudity, sexual activity, or sexually suggestive context, regardless of artistic intent.