All UsersContent Policies

Content Moderation Policy

Version 1.0Last updated 22 October 2026Effective 22 October 2026

Preamble

This content moderation policy describes when, why and how desioart reviews content, how unlawful content and content breaching our policies are dealt with, and what procedures apply to reports (notice and action), statements of reasons, trusted flaggers, orders from authorities and transparency reports. What is allowed and what is forbidden is governed by the AUP (Acceptable Use) and the community guidelines (Community); this document governs the procedure.

It implements:

Regulation (EU) 2022/2065 (Digital Services Act, DSA) — Article 6 (hosting), Articles 9 and 10 (orders from authorities), Articles 11 and 12 (points of contact), Article 14 (transparency of moderation), Articles 15 and 24 (transparency reports), Article 16 (notice and action), Article 17 (statement of reasons), Article 18 (reporting criminal offences), Articles 20 and 21 (complaints, out-of-court dispute settlement), Article 22 (trusted flaggers), Article 23 (measures against misuse), Article 27 (recommender systems), Article 28 (protection of minors)
Regulation (EU) 2021/784 (terrorist content online), Regulation (EU) 2019/1150 (P2B — restriction and termination towards creators), and the record-keeping and reporting obligations for child sexual abuse material (18 U.S.C. §§ 2257 and 2258A, so far as applicable)

It forms part of our terms of service (Terms section 19.1) and gives substance to Terms sections 3.4 (moderation tools) and 3.5 (recommender systems).

A voluntary commitment: for as long as we are exempt as a micro or small enterprise from individual DSA obligations (Articles 19 and 15(2) DSA), we apply the procedures described in this policy voluntarily, because we think them right. You can rely on that, and we treat the deadlines as binding targets. Should the exemption fall away, the procedures continue as statutory obligations.

1 Our approach to moderation

1.1 Basic principles

We are a hosting service provider and an online platform within the meaning of Article 3(g) and (i) DSA. Responsibility for the substance of content lies with the creators who publish it, and we are under no general obligation to monitor (Article 8 DSA). We nonetheless use the following, transparently and in proportion:

automated checks before publication (section 2.1): hash matching against known abuse material and against content already removed, malware checks, and a check on the consent of people taking part
human review (section 2.2): the release of public media, and reviews prompted by reports, by trusted flaggers, by orders from authorities, by escalations from our systems and by risk-based sampling
a notice and action procedure (section 3) with binding deadlines graded by seriousness
reasons for every measure (section 4) and an appeal (Appeals) decided by a human being
proportionate measures (section 5) — from a warning to a permanent ban, the latter only under the four-eyes principle

1.2 What we review

breaches of the AUP (Acceptable Use) — the zero-tolerance list, prohibited practices, the AI rules, account abuse, off-platform breaches
breaches of the community guidelines (Community) and of the rules for public media (Creator Terms section 5.5)
infringements of copyright and trade mark rights (DMCA) and impersonation
non-consensual intimate content (NCII Takedown)
content engaging the criminal law (abuse material, terrorism, trafficking, threats of violence, incitement to hatred)
missing or withdrawn consent from people taking part (Co-Performers, 2257 Statement)

1.3 What we do not review

the taste or the artistic quality of content
creators’ legitimate business decisions (prices, bundles, promotions), so far as they breach no consumer protection rules
private messages — we do not read along and we do not search message texts algorithmically. We look at one only on a report from someone involved, in appeal or complaint proceedings, or because the law obliges us, and every such access is logged. The exceptions (hash matching of every image and video file sent in messages, the consent checksum, AI persona chats) are described in Privacy section 2.6
political, religious and philosophical opinions, so far as they breach no rule on hatred, violence or terrorism
behaviour away from the platform with no connection to it

2 Our moderation tools

2.1 Automated checks

Tool	What is checked	What happens on a match	Provider and place
Hash matching against known child sexual abuse material (CSAM)	every uploaded image and every video (as stills) before publication or sending — posts, the media library, profile media, message attachments (Privacy section 2.10)	the file is never published, the original is preserved as evidence, the account is restricted provisionally, and an automatic case opens in child protection triage; a report goes to NCMEC and the authorities (section 6.4); no detailed reasons are given to the uploader	a specialist hash matching service (the hash is computed by us, the matching happens in an EU data centre); no human being needs to view anything
Matching against content already removed	perceptual hashes (PDQ, video hashes) of every upload against our block list	an upload of the same thing again is refused with a general notice	internal
NCII hash matching	hashes provided by the people affected through StopNCII	the upload is blocked and the team reviews it	StopNCII (hashes only)
Malware and file type checks	every file	quarantine, no upload	an open-source malware scanner, run by us
Checking public media are suitable for all audiences	profile picture, banner, intro video, promotional cover, public bio — before release to the public level	the item is not shown publicly until released, or is taken out of the public view and discovery (nothing deleted, no warning, no effect on earnings); an appeal is possible	reviewed by our own team in a queue of its own
Consent checks	before publication: does a consent exist for every person taking part who is marked? A count of people on the creator’s own device serves as a plausibility check	publication is blocked until consent exists; on a withdrawal, the content is hidden	internal; the count runs on your device using a library loaded from our servers — nothing is sent to anyone else
Report thresholds	the number of independent reports per item (section 3.6)	provisional hiding until a human being reviews it	internal
Comment word filter	comments against a list of prohibited terms	the comment goes into a moderation queue	internal
Detection of abusive reporting	patterns of false reports and co-ordinated mass reporting	loss of access to the reporting feature; mass reports treated as manipulation (section 3.7)	internal
Child protection triage	every report in the “minors” and “CSAM” categories: a rule-based assessment of metadata, hashes and verification status — no AI, and no viewing of content	immediate hiding, preservation of evidence, restriction of the account; classification as “report”, “fast review” or “noise” (closed as noise only where several signals agree and two people confirm)	internal

How far automation goes today: no classification of every post for nudity, violence or AI origin runs at present, and the release of public media is checked by people. We consider the combination of hash matching, verification of everyone involved, the reporting procedure and sampling proportionate, and it lets us work with less data. Should we bring in a classification tool later — to triage public media, say, or to moderate livestreams — we announce it here beforehand (section 11) and set out its purpose, accuracy, possible rate of error and safeguards in the transparency report (section 9.3, Article 15(1)(e) DSA). The line drawn in section 2.3 holds even then: such a tool triages, it does not decide.

Individual matching services (hash matching against abuse databases, NCII hashes) we switch on as soon as the provider in question has approved us; until then the other automated checks and human review apply.

Where a match needs a human being to review it, the content stays unpublished or hidden until the decision is made; a failure of the checking service never results in automatic release (fail closed).

2.2 Human review

Our compliance team handles reports, appeals, escalations from the systems, checks on public media and samples in separate queues; every decision is logged with its reasons, the reviewer’s identifier and the time.
No human being, no permanent ban: permanent bans, the retention of earnings from USD 1,000 upwards, freezing an account on a child protection suspicion, classifying a child protection report as unfounded, downgrading a trusted flagger and opening preserved evidence all require the four-eyes principle: two people in different roles, with two-factor confirmation.
While our team is small, an outside reviewer (a law firm or a compliance provider) carries out the second review; they are bound by contract to confidentiality and under Article 28 GDPR.
Reviewers never decide cases they are involved in themselves, and an appeal is examined by someone other than whoever made the original decision.
Reviewers are trained (protection of minors, NCII, trafficking, discrimination, protection from the strain of the work) and work to written guidelines; we publish the rules, not our internal thresholds.

2.3 The role of AI

Automated procedures serve to prepare the ground: hash matching, prioritisation, pattern recognition in reports. AI-based tools for classifying content we do not use at present; were they added, this division of roles would stay as it is. A human being decides about content and accounts, with two exceptions that we disclose: a hash match against verified abuse databases (section 2.1), and provisional hiding once a report threshold is reached (section 3.6). Both are identified in the statement of reasons as “automated means” (Article 17(3)(c) DSA) and can be reviewed (Privacy section 4.4).

3 Notice and action (Article 16 DSA)

3.1 Who can report?

every user
anyone without an account — anonymously if they wish
trusted flaggers under Article 22 DSA and our voluntary specialist partners (section 5)
authorities (section 6) and rights holders (DMCA)

3.2 How to report

Channel	For what
the “Report” icon on every post, comment and profile, and in every message	every category (section 3.3); the fastest route, because the content and its context are attached automatically
desioart.com/report	reports without an account (with a security check; up to 5 reports per 24 hours) — the categories are abuse material, minors, non-consensual content, unlawful content and threats
desioart.com/nci-takedown · ncii@desioart.com	intimate images without consent — the 24-hour procedure (NCII Takedown)
dmca@desioart.com	copyright and trade marks (DMCA)
abuse@desioart.com	anything else, by email
trafficking@desioart.com	confidential: coercion, exploitation, trafficking (Anti-Slavery)

Every mailbox acknowledges receipt automatically, with a case number.

3.3 What a report should contain

Information	Status
the content concerned (link, identifier — attached automatically in the reporting dialogue)	required
the category (section 3.5)	required
why the content is unlawful or breaks the rules	required for “anything else” and for infringements of rights; otherwise recommended
the reporting person’s name and email	voluntary — except for copyright reports, where the law requires it; for reports of abuse material, terrorism, trafficking or sexual services it is deliberately not required
a declaration that the information is correct to the best of your knowledge	confirmed in the form

A report that identifies content precisely enough gives us actual knowledge within the meaning of Article 16(3) DSA, and we examine it.

3.4 Anonymous reports and what we tell you

We examine anonymous reports just as carefully. Where you gave an email address, we acknowledge receipt without delay and tell you our decision, with reasons and a note of the remedies open to you (Article 16(5) DSA). We do not disclose your identity to the person reported; for copyright reports the statutory particularities apply (DMCA).

3.5 Categories, deadlines and thresholds

Reports are classified by seriousness. The deadlines are targets for reaching a decision, not merely for looking at it; the full list of categories is in Acceptable Use section 10.1.

Seriousness	Categories	Provisional hiding	Decision within
Critical	minors, abuse material, intimate images without consent, threats of violence, promotion of suicide or self-harm, terrorism, trafficking, cruelty to animals	at once, on the first plausible report	24 hours (child protection: without delay, target 1 hour)
Serious	harassment, hatred, doxxing, impersonation, other non-consensual content, illegal products, arranging sexual services, fraud and sextortion, prohibited sexual depictions, copyright	after 3 independent reports; copyright: on a valid notice	7 days
Medium	spam, off-platform selling, impermissible links, missing AI labelling, anything else	after 5 independent reports	14 days
Complex cases	—	as the situation requires	up to 30 days, with provisional measures and an interim response

Terrorist content we remove within one hour of an order under Article 3 of Regulation (EU) 2021/784. Orders to remove abuse material we carry out without delay.

3.6 Provisional hiding

Provisional hiding on reaching a threshold is an automated, provisional measure to protect the community; it is not a decision that a breach occurred. A human being reviews it within the deadline and makes the content visible again where there is no breach, with no consequence for the account. Repeated hidings with no breach behind them we treat as a sign of reporting abuse against the creator concerned.

3.7 Abusive and co-ordinated reporting (Article 23(2) DSA)

Anyone who repeatedly files reports that are plainly unfounded (as a guide: five rejected reports in 30 days) loses access to the reporting feature for 7 days after a warning, and for 30 days or permanently where it continues. Co-ordinated mass reporting — ten or more reports against the same item within an hour — we detect automatically; it does not make hiding any faster but is examined as an attempt at manipulation (Acceptable Use sections 5.1 and 10.6). Before we take away access to the reporting feature we hear from the person concerned, unless that is plainly unnecessary.

4 Statements of reasons (Article 17 DSA)

With every measure that restricts visibility, reach, monetisation or access to an account — removing or hiding content, removing it from discovery, a warning, a restriction, a temporary or permanent ban, a retention of earnings, refusing public media — the person concerned receives a statement of reasons as a notification of its own and permanently under “Account status” (/account-health). It sets out:

the measure, how long it lasts and where it applies
the facts and the content concerned
whether the decision rests on a report, an order from an authority or our own finding (the identity of a reporting person is not disclosed)
whether automated means played a part in detection or in the decision
the contractual clause breached (AUP, terms of service, community guidelines) or the legal provision, where the content is unlawful
the date and the case number
the remedies: an appeal (Appeals, 6 months), a complaint (Complaints), out-of-court dispute settlement (Article 21 DSA), the courts

We produce statements of reasons in German, English and French. We transmit each one, without personal data, to the European Commission’s DSA transparency database (Article 24(5) DSA).

Limited reasons: where offences against children are suspected, where evidence is being preserved for an ongoing investigation, or where an authority has ordered it, the reasons may be kept general so that investigations are not put at risk (Articles 17(1) and 18 DSA; Terms section 11.7). Where we remove something on an order from an authority we tell you, so far as we are allowed to, which authority issued it and what remedies you have against it (Article 9(5) DSA). For misleading commercial bulk content (spam) we may dispense with reasons (Article 17(2) DSA).

5 Trusted flaggers and specialist partners (Article 22 DSA)

5.1 Trusted flaggers under Article 22 DSA

Reports from organisations awarded trusted flagger status by a Digital Services Coordinator of an EU member state (the European Commission’s public list) are handled as a priority and without delay. Registering with us is not necessary; to speed things up, such organisations can have access to our reporting channel set up at legal@desioart.com.

5.2 Specialist partners we prioritise voluntarily

Regardless of formal status, we handle reports from the following bodies as a priority, usually within 24 hours, because experience shows their reports to be well founded:

Organisation	Country	Field
jugendschutz.net	Germany	protection of minors
INHOPE hotlines	international	abuse material
Internet Watch Foundation (IWF)	UK	abuse material
NCMEC	USA	abuse material, Take It Down
StopNCII.org / Revenge Porn Helpline	UK	intimate images without consent
HateAid	Germany	digital violence
Cyprus Safer Internet Center (CyberSafety)	Cyprus	protection of minors, hotline

We keep adding to this list.

5.3 Privileges and protection against misuse

Reports under sections 5.1 and 5.2 are taken into review first, treated as more plausible at the outset, and escalated straight to the head of the compliance team where necessary. Where it emerges that a trusted flagger is submitting a substantial number of imprecise or inadequately reasoned reports, we suspend the priority treatment after hearing from them and inform the competent Digital Services Coordinator (Articles 22(6) and (7) DSA).

6 Working with the authorities (Articles 9, 10 and 18 DSA)

6.1 Orders to act against content (Article 9 DSA)

Where a competent judicial or administrative authority orders us to act against a particular item of unlawful content, we check its competence, its form and the information Article 9(2) DSA requires (the legal basis, the reasons, a precise identification, the territorial scope, the remedies). Where an order is valid we carry it out without delay, tell the authority when and how, and inform the person concerned (section 4) unless the order forbids it. We keep the territorial scope to what is necessary (geoblocking rather than worldwide removal, where the order has national effect only).

6.2 Orders to provide information (Article 10 DSA) and law enforcement requests

We disclose information about users only on a lawful order from a competent authority, which identifies specifically the purpose, the legal basis and the users concerned — never on an informal request. We examine every request, confine the disclosure to what is necessary, and inform the person concerned unless the order forbids it or doing so would jeopardise the investigation.

What we are able to disclose at all: the verified name and date of birth (only for creators and co-performers who have passed an identity check), the username, the email address, the date and country of registration, payment and payout metadata (amounts, times, payment service providers — not full card numbers), content and messages, and sign-in times. IP addresses we hold in the clear only briefly, in server logs, and only hashed thereafter — attributing them retrospectively beyond a few weeks is technically impossible (Privacy section 7.1).

Urgent cases: where life or physical safety is in immediate danger, we may give the authorities the data needed to avert it without a formal order (Article 18 DSA; Article 6(1)(d) GDPR); every urgent disclosure is documented and reviewed afterwards.

Preservation requests: on a reasoned request we preserve content and data against deletion for up to 90 days, extendable, until an order arrives.

6.3 The point of contact for authorities (Article 11 DSA)

Orders and requests from authorities reach us at legal@desioart.com (subject “Law Enforcement Request” or “DSA Order”), in English, German and Greek. The point of contact for users (Article 12 DSA) is support@desioart.com (Terms section 20). We publish a guide for authorities at legal.desioart.com/law-enforcement.

6.4 Reporting criminal offences (Article 18 DSA; 18 U.S.C. § 2258A)

Where we learn of information giving rise to a suspicion of a criminal offence threatening people’s life or safety — abuse material, trafficking and threats of violence above all — we report it without delay to the competent law enforcement authorities of the member state concerned or of Cyprus and, for abuse material, to the NCMEC CyberTipline (target: within 24 hours). Confirmed hash matches are reported without any discretion. For content connected with the United Kingdom we inform the IWF, and for content connected with Germany the Bundeskriminalamt (target: 48 hours). We transmit only the data the report requires (Privacy section 5.3).

6.5 The competent authorities

Country or level	Bodies
Cyprus (our seat)	the Digital Services Coordinator; the Cyprus Police Cybercrime Unit; the Office of the Commissioner for Personal Data Protection
Germany	the Bundesnetzagentur (Digital Services Coordinator), the Bundeskriminalamt, the state media authorities and the KJM, jugendschutz.net
EU	Europol, the European Commission (the DSA transparency database), INHOPE
United Kingdom	the Internet Watch Foundation; NCA-CEOP for child protection cases connected with the UK
USA	NCMEC; the competent federal agencies for 2257 inspections

7 The measures, and what they do

Which measure we choose depends on seriousness, repetition, intent and harm (Article 23 DSA, Acceptable Use section 9). What each one does:

Measure	Effect	Duration
Removal from discovery (public media)	the item no longer appears in trending, search or recommendations; the profile stays reachable	until a permissible item replaces it
Hiding an item of content	the content is invisible to others and visible to you under “Account status”; the original is kept for as long as you can appeal	until the decision, or the appeal
A warning	a notice with reasons; it counts in weighing future breaches; warnings for child protection or intimate images without consent never expire	carries less weight as time passes without complaint
A restriction	particular features are blocked	24 hours to 30 days
A temporary ban	no access beyond reading, appealing and exporting your data; content invisible; balance frozen; subscriptions continue	24 hours, 7 days or 30 days; ends automatically
A permanent ban	the account is closed; content invisible; subscriptions ended and fans refunded pro rata; amounts connected with the breach retained and the rest of the balance handled under Creator Earnings section 10; circumvention protection active (Terms section 11.6)	indefinite; only after a four-eyes review
Locked out entirely	where a criminal matter is suspected or evidence is being preserved: no sign-in; appeals and data access by email (appeals@ and privacy@)	until the matter is clear

Retention after a ban: content in permanently banned accounts is kept at least for as long as an appeal is possible (6 months) and deleted after that, unless the preservation of evidence or a statutory retention duty prevents it; 2257 records and verification data are kept under Privacy section 7.1. Where a ban is lifted on appeal, we restore your content, your reach and your balance.

8 Recommender systems and reach (Article 27 DSA)

The main parameters of our ordering (the subscription feed chronologically; trending and new by activity score; recommended creators by similarity; the top-creator badge; the promotions page) are described in Terms section 3.5. What matters for moderation:

Moderation decisions affect reach: hidden content, restricted accounts and public media not suitable for all audiences are taken out of trending, search and recommendations, and we say so in the statement of reasons (“removal from discovery”).
No covert throttling of reach (“shadowbanning”): every restriction on visibility comes with reasons (Article 17(1)(a) DSA).
Controls for creators: creators can take themselves out of search (“not discoverable”), limit comments and block countries; those settings are not moderation measures.
Protection of minors before reach: unverified visitors see only checked media suitable for all audiences (Age Verification section 1).
For recommendations we use no special categories of personal data (Article 26(3) DSA) and no profiles of minors (Article 28(2) DSA — minors have no access).

9 Records, preservation of evidence and transparency

9.1 Records of decisions

Every statement of reasons, every report and every appeal is recorded with a case number, a timestamp, the reviewer’s identifier and the outcome (kept for 5 years after the case closes, Privacy section 7.1). We transmit statements of reasons to the DSA transparency database automatically, and the summaries are publicly visible there, without personal data.

9.2 Preservation of evidence

Where abuse material, intimate images without consent or trafficking is suspected, or an authority asks us to, we preserve the content and the associated data immutably (WORM storage, encrypted, accessible only with four-eyes approval and a logged purpose) for 5 years, or until the proceedings end (a legal hold). Preserved evidence is never used for anything else.

9.3 The transparency report (Articles 15 and 24 DSA)

We publish a transparency report annually (first for our first full year of operation) at desioart.com/transparency, covering:

the number of orders from authorities under Articles 9 and 10, by member state and type of content, with the average handling time
the number of reports under Article 16, by category, by origin (users, trusted flaggers) and by outcome, the proportion handled by automated means, and the average handling time
the number and kind of moderation measures taken on our own initiative, categorised by violation, detection method and restriction applied (Article 15(1)(c)), and the automated means we use with their precise purposes, indicators of their accuracy and possible rate of error, and the safeguards applied (Article 15(1)(e))
the number of appeals under Article 20, their outcome, the average time to decide, and the number of decisions reversed
the number of dispute settlement proceedings under Article 21 and their outcome
the number of restrictions under Article 23 (manifestly illegal content, unfounded reports, unfounded complaints)
reports to authorities and to NCMEC, in aggregate

In addition we publish the average monthly number of active users in the EU every six months (Article 24(2) DSA) and give it to the Digital Services Coordinator on request. The report contains no identifying data. It is published in machine-readable and easily accessible form.

10 How this fits with our other policies

Matter	The detailed procedure
What is allowed and what is forbidden	Acceptable Use, Community, Creator Terms section 5.5
Copyright and trade marks, impersonation	DMCA
Intimate images without consent	NCII Takedown
Suspected minors	Age Verification section 10
Coercion, exploitation, trafficking	Anti-Slavery
Complaints about the platform or about other users	Complaints
Appeals against moderation decisions	Appeals
People taking part, and their consent	Co-Performers, 2257 Statement
Data processing in moderation and the preservation of evidence	Privacy sections 2.10, 2.14, 5.3 and 7.1
Restriction and termination towards creators (P2B)	Creator Terms sections 12 and 15

11 Updates

Where a change is substantial we tell you 30 days beforehand (creators at least 15 days, Article 3 of Regulation (EU) 2019/1150); changes needed to avert an unforeseen danger, or required by law, take effect at once. New automated tools we announce here before using them (Article 14 DSA). The current version and a record of the changes are at legal.desioart.com/content-moderation.

12 Contact

Matter	Channel
Reporting content	the “Report” icon · desioart.com/report · abuse@desioart.com
Intimate images without consent (24 h)	desioart.com/nci-takedown · ncii@desioart.com
Copyright and trade marks	dmca@desioart.com
Coercion and trafficking (confidential)	trafficking@desioart.com
Appealing a decision	“Account status” in your account · appeals@desioart.com
Complaints	complaints@desioart.com
Authorities, orders, trusted flaggers	legal@desioart.com
The transparency report, press	legal@desioart.com

Postal address: 1 Apriliou 52, 7600 Athienou, Larnaca, Cyprus