YOUR PRIVACY
The Full Score
How we handle your personal data on Pash — composed so everyone can follow along
Think of your personal data like a song you wrote.
This notice tells you who gets to play it, how, and when.
You stay the composer. Always.
This notice satisfies Article 13 UK GDPR and Article 13 EU GDPR (Regulation 2016/679).
Tracklist
Track 1 — Meet the Band (Who We Are)
Every song needs a band, and every platform needs someone responsible for your data. That's us. In legal language, we're the “data controller” — the ones who decide how your personal information is used.
Sheet music reference: Art. 13(1)(a) — The controller must identify themselves and provide contact details, including any representative.
| Role | Details |
|---|---|
| Band name (company) | Muse Social Network Ltd |
| Studio address | 1 COLDBATH SQUARE, LONDON, ENGLAND EC1R 5HL |
| Company number | 166880335 |
| Drop us a line | hello@pash-app.com |
Track 2 — The Sound Engineer (Our Data Protection Officer)
Every good studio has a sound engineer making sure everything runs cleanly. Our Data Protection Officer (DPO) does the same for your data. If something sounds off, reach out to them.
Sheet music reference: Art. 13(1)(b) — Contact details of the DPO.
| DPO name | Giuseppe Matteo Miroddi |
| privacy@downloadmuse.com |
Track 3 — The Setlist (What We Do With Your Data and Why)
A setlist tells you which songs will be played and in what order. Here's our setlist — every way we use your data, and the legal reason (called a “lawful basis”) behind each one.
Sheet music reference: Art. 13(1)(c) — Purposes and lawful basis for each processing activity.
3.1 Opening Act — Your Account
Legal reason: Contract — we need this to give you the service you signed up for.
When you join, we use your username, email, and profile info to set up your stage — your account. This lets you post, message, follow, and use all the features.
3.2 The DJ Set — Your Personalised Feed
Legal reason: Legitimate interest — we have a real, fair reason for doing this.
Like a DJ reading the room, we look at what you listen to (interact with) and tune your feed so you see more of what you enjoy. We think this makes the experience better for you.
Not feeling the mix? You can switch to a chronological feed any time — see Track 8.
3.3 The Sponsors — Advertising
Legal reason: Consent — only if you give us permission.
If you say yes, we use things like what you browse, your device type, and your rough location to show ads that might actually interest you (instead of random noise). We also check if those ads worked.
You can pull the plug any time — go to settings or our cookie tool. Changing your mind doesn't undo anything that already happened legally.
3.4 The Bouncers — Platform Safety
Legal reason: Legitimate interest + legal obligation.
Nobody wants a bad act ruining the show. We check reports to spot rule-breaking, illegal material, spam, and bullying. We follow the Online Safety Act 2023 (UK) and Digital Services Act (EU) — think of them as venue safety regulations.
3.5 The Mixing Desk — Analytics
Legal reason: Legitimate interest.
We look at overall patterns — which features are hits, which are flops — so we can improve the platform. This data is grouped together or disguised; we're reading the crowd, not tracking one person.
3.6 Backstage Pass — Messages From Us
Legal reason: Contract (essential alerts) or consent (marketing).
We send you need-to-know messages (security alerts, rule changes). If you opt in, we may also send marketing. You can mute marketing any time.
3.7 The Contract With the Venue — Legal Obligations
Legal reason: Legal obligation.
Sometimes the law requires us to keep or share data — like when a court or regulator asks. Think of it as the venue's licence conditions: non-negotiable.
3.8 Hidden Tracks — Sensitive Information
We don't ask for sensitive info (race, religion, health, sexuality). But sometimes what you post or do on the app might reveal these things — like hidden tracks on an album you didn't expect.
If that happens, we only use it with your clear agreement, or because you chose to put it on your public profile for everyone to hear.
Sheet music reference: Art. 9 — Special category data. DPIAs conducted where sensitive inferences affect profiling or advertising.
Track 4 — Our Reasons to Play (Legitimate Interests)
When we say we have a "legitimate interest," it means we have a real, honest reason — and we've checked it doesn't drown out your rights. Here's what we're playing for:
Sheet music reference: Art. 13(1)(d) — Where the controller relies on legitimate interests, they must tell you what those interests are.
- Tuning your feed: showing you content that matches your taste
- Keeping the venue safe: stopping abuse, fraud, spam, and illegal content
- Improving the sound: understanding how people use features to make them better
- Guarding the equipment: protecting the platform and your data from hackers
We've written these reasons down in detail (a Legitimate Interests Assessment). Ask our DPO for a copy.
Track 5 — The Guest List (Who Hears Your Data)
Your data isn't broadcast to everyone. Here's who's on the guest list and why:
Sheet music reference: Art. 13(1)(e) — Recipients or categories of recipients.
- Cloud providers (the studio): they store and run everything behind the scenes
- Analytics partners (the producers): help us understand overall trends — no names attached
- Ad partners (the promoters): only if you agreed to personalised ads
- Moderation helpers (the security crew): help us review flagged content
- Payment processors (the ticket office): if you buy something in the app
- Authorities (the licensing board): police or regulators, only when the law requires it
- Our advisors (the managers): lawyers and auditors, bound by confidentiality
Everyone on this list has signed an agreement to protect your data.
Track 6 — World Tour (International Data Transfers)
Sometimes your data travels outside the UK or Europe — for example, some companies helping us are based in the US. When your data goes on tour, we make sure it travels with protection.
Sheet music reference: Art. 13(1)(f) — Transfers to third countries, safeguards, and how to get copies.
6.1 Leaving the UK
- Approved destinations: countries the UK government says protect data well enough
- UK-US Data Bridge: a special arrangement for certified US companies
- UK IDTA / Addendum: a legal contract forcing the other side to protect your data
6.2 Leaving the EU
- EU adequacy decisions: EU-approved safe countries
- EU-US Data Privacy Framework: EU version of the US bridge
Track 7 — How Long the Record Plays (Retention)
We don't keep your data on repeat forever. Here's how long each track plays:
Sheet music reference: Art. 13(2)(a) — Retention periods or criteria for determining them.
| Your data | How long it plays | Why |
|---|---|---|
| Account details | While your account is open + 30 days after you close it | Needed to run your account |
| Your posts and content | Until you delete them or leave | You control the tracklist |
| Private messages | Until you delete them or leave | Part of the service |
| What you interact with | months | Powers your feed; then deleted |
| Ad data | 14 months | Only while you've given permission |
| Moderation records | 5 years after resolved | Safety and legal reasons |
| Technical logs | 12 months | Security |
| Rough location | Until you delete them or leave | Only with your permission |
When the track ends, we securely erase it or make it completely anonymous — like wiping the master tape.
Track 8 — Your Remix Rights
This is your music. The law gives you powerful rights over it. Use them any time by emailing privacy@downloadmuse.com or through your in-app privacy settings.
Sheet music reference: Art. 13(2)(b) — Rights of access, rectification, erasure, restriction, objection, and portability.
8.1 The basics
- Listen back (access): Ask us what data we have about you.
- Re-record (rectification): If something's wrong, ask us to fix it.
- Hit delete (erasure): Ask us to erase your data in many situations.
- Press pause (restriction): Ask us to stop using your data while we sort something out.
- Export your tracks (portability): Get your data in a standard digital format and take it to another platform.
8.2 Stop the music
Whenever we use your data on a "legitimate interest" basis (Track 4), you can object at any time. We'll stop unless we have a very strong reason that outweighs your rights.
Sheet music reference: Art. 13(2)(b) — Right to object (Art. 21), including to profiling.
8.3 Revoke the licence
If we asked your permission (consent), you can take it back any time. This doesn't make anything we did before illegal — it just stops the music going forward.
Sheet music reference: Art. 13(2)(c) — Right to withdraw consent (Art. 7(3)).
Response time: We'll get back to you within one month. Complex requests may take up to three months total — we'll tell you.
Track 9 — Feedback to the Critics (Complaints)
If you think we hit a wrong note with your data, you can complain to a data protection authority.
Sheet music reference: Art. 13(2)(d) — Right to lodge a complaint with the supervisory authority.
9.1 UK — The ICO
- Website: ico.org.uk
- Phone: 0303 123 1113
- Address: ICO, Wycliffe House, Water Lane, Wilmslow, Cheshire SK9 5AF
9.2 EU — Your local authority
Complain to the data protection authority where you live, work, or where the problem happened.
Track 10 — Must-Play vs Freestyle (What Data Is Required)
Sheet music reference: Art. 13(2)(e) — Whether providing data is obligatory, and consequences of not providing it.
10.1 Must-play (required)
To create an account, we need your email, a username, and your date of birth. No ticket, no entry. This is part of our contract with you.
10.2 Freestyle (optional)
Profile photos, bios, posts — all optional. You can enjoy the show without them, but some features work better with them.
10.3 Venue rules (legal requirement)
Sometimes the law says we must collect specific data (e.g. age checks). We'll tell you at the door.
Track 11 — The Algorithm (Automated Decisions and Profiling)
Sheet music reference: Art. 13(2)(f) — Automated decision-making including profiling: logic involved, significance, and consequences.
11.1 The playlist algorithm
Like a streaming service building you a playlist, our system looks at what you read, like, share, and how long you linger on posts. It then ranks other content by how likely you are to enjoy it. This is profiling — but it doesn't produce legal effects. It just shapes your feed.
11.2 Automated moderation
We might use automated tools to spot rule-breaking content (spam, illegal material). Think of it as an auto-tune filter — it catches obvious problems. But a real person reviews before we take serious action on your account.
If an automated decision significantly affects you, you can:
- Ask a human to review it
- Tell your side of the story
- Challenge the decision
Sheet music reference: Art. 22 UK GDPR (with Art. 22C per DUAA 2025) and Art. 22 EU GDPR — safeguards for significant automated decisions.
11.3 How it works — plain version
Our algorithm scores posts based on: what you've interacted with, who you follow, what you liked, how much time you spent, and when the posts happened in chronological order. Higher-scoring posts appear first. No single signal controls everything — it's a full mix, not a solo.
Track 12 — Bonus Tracks (Extra Things to Know)
12.1 Young musicians (under 18)
We know younger people use this platform. Extra protections apply:
- Privacy settings default to maximum protection for younger users
- No personalised ads for users we know are under 18
- Privacy information in plain language — like this notice
If you're under the age of digital consent (13 in the UK; 13–16 across EU countries), a parent or guardian needs to agree before we use your data based on consent.
Sheet music reference: ICO Age Appropriate Design Code; EDPB guidance on children's data; European Commission DSA guidelines on minors.
12.2 Safety checks (DPIAs)
For our riskiest processing, we've done detailed impact assessments (DPIAs) covering:
- Content recommendation algorithm
- Automated moderation
- Ad profiling
- Processing data of under-18s
- Anything involving sensitive inferences
Ask our DPO for summaries.
12.3 Setlist changes
If we change this notice significantly, we'll notify you through the app or email before changes kick in.
12.4 New songs (further processing)
If we ever want to use your data for a brand-new purpose — a track we haven't announced yet — we'll tell you first and explain it before we press record.
Sheet music reference: Art. 13(3) — Prior notice required before further processing for a new purpose.
Track 13 — Encore (Contact Us)
Questions? Concerns? Want to use your rights? Here's how to reach us:
- General privacy: privacy@downloadmuse.com
Remember: it's your data, your music, your rights. Don't be shy about using them.
Safety & Moderation
Standards against child sexual abuse and exploitation (CSAE), and moderation policy
Purpose of the moderation policy [a]
The Online Safety Act 2023 sets out a series of obligations that Muse shall be bound by regarding the control of the content inside the platform.
OSA-Compliant Prohibited Word List — English
Version: 1.0 — May 2026 Purpose: Automated keyword filter for social media content moderation Architecture: Word match → action per severity tier. User reporting supplements automated detection.
SEVERITY TIERS
| Tier | Action | Response time |
|---|---|---|
| TIER 1 — BLOCK | Auto-reject comment. Log for audit. Report to authorities where required (CSEA → NCA/IWF) | Immediate after three reporting |
| TIER 2 — HOLD | Auto-hold comment (invisible to public). Queue for human moderator review | < 24 hours |
| TIER 3 — FLAG | Comment posted but flagged internally. Moderator reviews in batch | < 72 hours |
TIER 1 — IMMEDIATE BLOCK
1A. Child Sexual Exploitation & Abuse (OSA Schedule 6)
Note: CSEA keyword lists are classified and maintained by law enforcement (NCA, IWF, NCMEC). Platforms should:
- Subscribe to the IWF URL List and hash list
- Use PhotoDNA or similar perceptual hash-matching
- Contact the IWF (iwf.org.uk) for access to their keyword advisory list — this is not published publicly for operational security reasons
General flaggable patterns (not exhaustive):
- Any sexual term combined with terms indicating minors (age references, school-related terms, child-related terms)
- Known CSEA-community coded terms (evolve rapidly — maintained via IWF/NCA feeds)
Legal obligation: Report confirmed CSEA content to the National Crime Agency immediately.
1B. Direct Terrorism Incitement (OSA Schedule 5)
Terms/phrases — block when used as direct incitement or glorification:
- Direct calls to commit terrorist attacks
- Glorification phrases praising specific terrorist acts or perpetrators
- Recruitment slogans for proscribed organisations
Reference list: UK Home Office publishes the list of proscribed organisations. Names of these organisations used in glorification/recruitment context → Tier 1. See: https://www.gov.uk/government/publications/proscribed-terror-groups-or-organisations--2
1C. Direct Death Threats (OSA s.181)
Phrases expressing direct intent to kill or cause serious harm to identifiable person:
- "I will kill you"
- "I'm going to murder you"
- "you're dead" (when directed at specific person as threat)
- "I'll find you and end you"
- "I will slit your throat"
- "I'm going to stab you"
- "I'll put a bullet in you"
- "you won't survive this" (threat context)
- "I will burn your house down"
- "your family will pay"
- Variants with misspellings, spacing, special characters
TIER 2 — HOLD FOR REVIEW
2A. Racial & Ethnic Slurs (OSA Schedule 7 — Hate offences; s.62(2)(a))
The following are commonly recognised racial and ethnic slurs in English. Include all known variant spellings, leetspeak substitutions (e.g., replacing letters with numbers), and deliberate misspellings:
- Anti-Black slurs: "ngger", "ngga", "coon", "darkie", "jigaboo", "sambo", "spook" (racial context), "tar baby", "pickaninny", "golliwog"
- Anti-Asian slurs: "chnk", "g**k", "pki", "slant-eye", "yellow" (racial slur context), "ching chong", "zipperhead"
- Anti-White slurs: "cracker" (racial context), "honky", "gringo" (derogatory context)
- Anti-Latino slurs: "sp*c", "wetback", "beaner"
- Anti-Arab/Middle Eastern slurs: "raghead", "towelhead", "sand n*gger", "camel jockey"
- Anti-Roma slurs: "gy*po", "pikey"
- Anti-Indigenous slurs: "redskin" (racial context), "savage" (racial context)
- Anti-Irish slurs: "fenian" (sectarian context), "taig"
- General dehumanising terms: "subhuman", "vermin", "cockroach", "animal" — when directed at ethnic/racial groups
Implementation note: Many of these words have non-slur meanings in other contexts (e.g., "cracker" = food, "spook" = ghost/spy). This is why Tier 2 = hold for human review, not auto-block.
2B. Homophobic & Transphobic Slurs (OSA Schedule 7 — Hate offences; s.62(2)(d)(f))
- "fggot" / "fg" (slur context, not food meaning)
- "d*ke" (slur context)
- "queer" — context-dependent: reclaimed by many LGBTQ+ people, slur when used with hostile intent
- "tranny"
- "shemale"
- "he-she", "it" (when deliberately used to dehumanise trans person)
- "groomer" (when falsely targeting LGBTQ+ individuals)
- "sodomite" (slur context)
- "batty boy" / "batty man"
- "bender"
- "poof" / "poofter"
2C. Disability Slurs (OSA s.62(2)(e))
- "retard" / "retarded" (directed at person)
- "spastic" / "spaz" (directed at person)
- "cripple" (slur context)
- "vegetable" (dehumanising disabled person)
- "window-licker"
- "mong" / "mongoloid"
2D. Religious Hate (OSA Schedule 7; s.62(2)(b))
- Antisemitic slurs: "k*ke", "yid" (slur context), "Christ-killer", "hook-nose"
- Islamophobic slurs: "Paki" (already listed above, dual category), "jihadi" (when used as blanket slur for Muslims), "goatf***er"
- Holocaust denial phrases: "holohoax", "6 million lie", "didn't happen"
- Genocide glorification/denial for any group
2E. Violence Promotion & Glorification (OSA s.62(4)(6))
- "kill all [group]"
- "gas the [group]"
- "[group] should be exterminated"
- "[group] deserve to die"
- "ethnic cleansing"
- "race war" (incitement context)
- "day of the rope"
- Glorification of mass shooters/attackers by name + praise
- "based" + violence reference (far-right coded approval)
2F. Suicide & Self-Harm Promotion (OSA s.61(3)(4), s.184)
Terms that promote or instruct — NOT terms used in help-seeking or recovery:
- "how to kill yourself"
- "best way to die"
- "suicide methods"
- "painless death methods"
- "pro-suicide"
- "catch the bus" (suicide community euphemism)
- "CTB" (abbreviation: "catch the bus")
- "end it all"
- "methods thread"
- Self-harm encouragement: "cut deeper", "you should cut", sharing methods/tools for self-injury
- References to specific lethal methods + encouragement
Critical context distinction: "I want to kill myself" = help-seeking → route to support resources, NOT block. "Here's how to kill yourself" = instruction → Tier 2 hold.
2G. Eating Disorder Promotion (OSA s.61(5))
- "pro-ana" (pro-anorexia)
- "pro-mia" (pro-bulimia)
- "thinspiration" / "thinspo"
- "bonespo"
- "meanspo" (mean inspiration — bullying into eating less)
- "purge tips"
- "laxative abuse tips"
- "goal weight" + extremely low BMI references
- "nothing tastes as good as skinny feels"
- "fasting challenge" (extreme/competitive context)
- "reverse thinspiration"
- "ED goals"
Context note: Recovery communities, nutrition advice, clinical content = lawful. Flag pattern = promotional/instructional + extreme weight loss.
2H. Intimate Image Abuse / Cyberflashing (OSA s.66A, s.66B SOA 2003)
- "I'll leak your nudes"
- "sending your pics to everyone"
- "revenge porn"
- "I have your photos" (threatening context)
- "expose you" (intimate image threat context)
- Unsolicited sexual imagery (detected via image classifier, not keyword)
2I. Harassment & Stalking Patterns (OSA Schedule 7)
Not single words but patterns:
- Repeated @mentions of same user with hostile content (volume-based detection)
- Doxxing: posting addresses, phone numbers, workplace details of private individuals
- "I know where you live"
- "I've found your address"
- "I'm outside your house"
- Coordinated pile-on indicators (multiple accounts, same target, short timeframe)
2J. Fraud & Scams (OSA Schedule 7)
- "Send me your bank details"
- "DM me for investment opportunity"
- Impersonation of official accounts (brand name + "official" + solicitation)
- "Guaranteed returns"
- "Double your crypto"
- Phishing link patterns
2K. Drug Dealing (OSA Schedule 7)
- "DM for [substance name]"
- "selling [controlled substance]"
- Common emoji codes used in drug sales (evolve rapidly — maintain via law enforcement feeds): e.g., specific plant/pill/snow emojis in sales context
- Slang for controlled substances when combined with sales language ("packs", "plugs", "re-up")
2L. Weapons Offences (OSA Schedule 7)
- "selling [firearm type]"
- "buy guns UK"
- "3D printed gun files"
- "ghost gun"
- "knife for sale" (illegal weapon context)
- Instructions for manufacturing weapons
TIER 3 — FLAG FOR CONTEXT REVIEW
3A. Profanity (General — Context-Dependent)
Common profanity alone is NOT illegal under OSA. Flag only when combined with targeting patterns:
- "fuck", "shit", "bitch", "bastard", "cunt", "dick", "arse/ass", "wanker", "twat", "bollocks", "piss off", "slag", "whore", "slut"
Action: Flag only when: directed at specific user + repeated + abusive pattern. General profanity in non-targeted context = no action required under OSA.
3B. Mild Hostile Language
- "idiot", "moron", "stupid", "loser", "ugly", "fat" (body-shaming), "die" (casual), "trash", "worthless", "pathetic", "freak", "weirdo", "creep"
Action: Flag only when part of sustained bullying pattern (s.62(5)(12): campaign of mistreatment). Single instances = no action needed.
3C. Political / Controversial Terms
These are NOT prohibited by OSA but may appear in reports:
- Terms around immigration debate, political figures, policy disagreements
- Do NOT auto-moderate. OSA protects political speech. Only act if combined with Tier 1 or Tier 2 content.
3D. Dangerous Challenges (OSA s.62(8))
- "[challenge name] challenge" (monitor trending challenge names — maintain dynamic list)
- "try this dare"
- "I dare you to"
- Context: challenges involving physical risk, substance ingestion, dangerous stunts
IMPLEMENTATION NOTES
Pattern Matching Rules
- Case-insensitive matching on all terms
- Leetspeak substitution — detect common letter→number swaps: a→4, e→3, i→1, o→0, s→5/$, t→7
- Space/character insertion — detect "n i g g e r", "n.i.g.g.e.r", "n*gger" etc.
- Deliberate misspelling — maintain variant list, update monthly
- Unicode homoglyphs — detect Cyrillic/Greek character substitution (е→e, о→o, etc.)
- Zero-width characters — strip before matching
- Emoji combinations — maintain evolving code list via law enforcement advisories
False Positive Mitigation
- Tier 2 and 3 = human review, never auto-delete
- Allowlist common false positives: "cracker" (food), "spook" (film/ghost), "faggot" (UK meatball dish), "niggardly" (unrelated etymology but commonly flagged)
- Quoting/discussing slurs in educational or reporting context ≠ using them — moderators must distinguish
- News articles, academic discussion, counter-speech = protected
User Reporting Integration
- Report button on every comment
- Report categories map to OSA categories: Hate speech, Threats, CSEA, Terrorism, Self-harm/suicide, Fraud/scam, Bullying, Other
- User reports feed same moderation queue
- Reported content gets priority review regardless of keyword match
Record-Keeping (Ofcom requirement)
- Log every moderation action: keyword triggered, tier, action taken, reviewer (if human), timestamp, outcome
- Retain logs per Ofcom retention guidance
- Produce transparency reports
Update Cadence
- Weekly: Review new evasion patterns, add variants
- Monthly: Review false positive/negative rates, tune keyword list
- Quarterly: Full review against Ofcom bulletins, IWF advisories, NCA updates
- On-demand: When Ofcom issues new codes or guidance