Social Media — Category Research Report

What you share, who you connect with, what you see — owned by companies that sell your attention.

What you share, who you connect with, what you see — owned by companies that sell your attention. This is the landscape, the data, and the opportunity.


The Landscape

Social media is a $250B+ annual revenue market. The dominant players (updated with 2025 actuals and early 2026 data):

PlatformOwnerMonthly Active UsersRevenue ModelAnnual Revenue (2025)
FacebookMeta3.2BAdvertising$200.97B (Meta total)
InstagramMeta2-3BAdvertising(included in Meta)
WhatsAppMeta2-3BBusiness API + emerging ads(included in Meta)
YouTubeGoogle/Alphabet~2.5BAdvertising + Premium ($14/mo)~$35B
TikTokTikTok USDS JV (US) / ByteDance (global)1.9BAdvertising + TikTok Shop~$28.4B total; ~$33B ad revenue
ThreadsMeta450MAdvertising (global rollout 2025)~$8B (est. 2025)
X (Twitter)xAI/Musk~557M (disputed; ~132M mobile DAU)Advertising + Premium ($8/mo)~$2.5-2.9B
SnapchatSnap Inc.946M MAU / 474M DAUAdvertising + Snapchat+ ($4/mo)$5.93B
LinkedInMicrosoft1.2B members / ~310M activeAdvertising + Premium ($30-60/mo)~$20B+ run rate (Q4 2025: $5B+)
RedditReddit Inc. (public, IPO March 2024)1.36B monthly / 121M DAUAdvertising + Premium + Data Licensing$2.2B
BlueskyBluesky PBC40.2MNot yet monetizedN/A
Mastodon/FediverseDecentralized (open source)~1-1.5M active / ~12M registeredDonations / self-hostedN/A

Sources: Meta Q4 2025 earnings (Jan 28, 2026): Meta Investor Relations, CNBC. TikTok users: DemandSage, Buffer. TikTok revenue: WARC, Cropink. X/Twitter: Business of Apps, RecurPost. Snapchat Q4 2025: Snap Investor Relations. LinkedIn: Cognism. Reddit Q4 2025: BusinessWire. Bluesky: Backlinko, Sprout Social. Threads: Backlinko, Social Media Today. Mastodon: FediDB, ThinkImpact.

Revenue comes almost entirely from advertising. Users pay with attention and data, not money. The user is the product — this is not metaphor, it is the literal business model.

Meta Platforms (Facebook, Instagram, WhatsApp, Threads)

Full Year 2025 Results (reported January 28, 2026):

  • Total revenue: $200.97 billion (+22% YoY) — a record year
  • Q4 2025 revenue: $59.89 billion (+24% YoY)
  • Family of Apps revenue: $58.9B in Q4 (+25% YoY)
  • Reality Labs revenue: $2.27B for full year (operating loss of ~$6B/quarter)
  • Net income: $22.8B in Q4 (+9% YoY)
  • EPS: $8.88 Q4 (beat forecast of $8.19 by 8.4%)
  • Average price per ad increased 9% YoY for the full year
  • Family of Apps daily active people: 3.58 billion (December 2025, +7% YoY)
  • Family of Apps monthly active people: 3.98 billion
  • Headcount: 78,865 employees (+6% YoY)
  • 2026 guidance: Q1 revenue $53.5-56.5B; full year capex $115-135B (AI infrastructure)
  • Mark Zuckerberg: "I'm looking forward to advancing personal superintelligence for people around the world in 2026."

Platform-specific user counts:

  • Facebook: ~3.07-3.2B MAU
  • Instagram: 2-3B MAU (sources vary; Meta does not break out individually)
  • WhatsApp: 2-3B MAU; ~130 billion messages per day
  • Threads: 450M+ MAU as of January 2026 (up from 320M in January 2025, 200M in September 2024); ~137-150M DAU; ads rolling out globally; projected $8B revenue in 2025, $11.3B in 2026

Source: Meta Q4 2025 Earnings, CNBC, Yahoo Finance, Backlinko Facebook Users, Backlinko Threads Users

TikTok

Users (2025-2026):

  • Global MAU: ~1.9 billion (early 2026), up from 1.59B in early 2025
  • US users: ~136-153 million (sources vary)
  • EU users: 169M MAU (official DSA report, H1 2025)
  • Daily active users: ~1.12 billion globally
  • Average daily usage: 52 minutes (US), 95 minutes (global)

Revenue:

  • 2025 total revenue (estimated): ~$28.4 billion
  • 2025 global ad revenue (projected): ~$33 billion (+40% YoY)
  • 2026 global ad revenue forecast: $34.8-44B (depending on source)
  • US ad revenue 2025: $11.8B; 2026 forecast: $13.4-14.5B
  • TikTok Shop (e-commerce): projected to exceed $20B in 2026

US Ban and Ownership Resolution:

  • Supreme Court upheld divest-or-ban law (PAFACA) on January 17, 2025
  • TikTok briefly went dark January 18-19, 2025; restored after Trump executive order
  • Four successive executive order extensions through 2025
  • Deal signed December 18, 2025; finalized January 22, 2026
  • New entity: TikTok USDS Joint Venture LLC
  • Investors: Oracle, Silver Lake, and MGX (~15% each), plus Dell Family Office, Vastmere, Alpha Wave Partners, existing ByteDance investors
  • ByteDance retains 19.9% minority stake (below 20% legal cap)
  • ByteDance's recommendation algorithm to be copied and retrained on US data only
  • Oracle provides data hosting and code review
  • 7-member, majority-American board of directors
  • Critics note: ByteDance entities still run advertising, e-commerce, and marketing for TikTok USDS; operational relationship not fully severed
  • Post-deal (Jan 22, 2026): TikTok updated terms and conditions expanding demographic data collection and precise location tracking

Source: ABC News, TechCrunch, CBS News, DemandSage, WARC

X (formerly Twitter)

Users:

  • ~557M MAU worldwide (mid-2025 estimate), though figures are disputed and vary widely (421M-619M depending on source and time period)
  • ~132M mobile daily active users (June 2025, per SimilarWeb) — slowly declining
  • US: 105M+ accounts
  • Lost ~33 million monthly users since Musk acquisition
  • UK filings (January 2026): 58% revenue plummet in UK operations

Revenue:

  • Estimated 2025 total revenue: ~$2.5-2.9B (down from $5.08B peak in 2021)
  • Ad revenue: ~$2.26B in 2025 (projected partial recovery of 16.5%)
  • X Premium subscriptions: ~$200M/year
  • X/xAI merger expected to bring $500M by end of 2025, rising to $2B in 2026
  • Revenue per user: ~$5.36 (vs. Facebook's $40+)
  • Valuation: bottomed at $9.4B (September 2024), recovered to $33B (March 2025), internal employee plan valuation at $44B

Workforce: Reduced from ~7,500 to ~2,840 employees by early 2025

Major Policy Changes (January 2025): Meta followed X's lead by ending third-party fact-checking in favor of community notes. X's model — where users provide context notes on posts — has become the template for the industry, despite studies showing limited effectiveness at reducing engagement with misinformation.

Source: Business of Apps, RecurPost, Charle Agency, The Frank Agency

Bluesky

Users: 40.2M total users as of late 2025 (302% increase from 10M in September 2024); ~3.09-3.5M daily active users Growth rate: Slowed from ~5M/month at peak to ~1.4-1.6M/month by mid-2025; ~17,280 new users daily Funding: $23M total ($8M seed July 2023; $15M Series A led by Blockchain Capital, October 2024) — approximately $0.84/user Team: 25 full-time employees, 100 contractors (content moderation) Revenue: Not yet monetized; plans for domain registration and potential subscriptions

AT Protocol Status:

  • IETF Internet Draft published September 2025; working group charter published January 2026
  • Architecture: Personal Data Servers (PDS), relays, App Views — "shared heap" rather than message-passing federation
  • 40,000+ algorithmic feeds available (user-composable)
  • Key advantage over ActivityPub: account portability (change servers without losing identity, data, or social graph)
  • Bridgy Fed software enables crossposting between AT Protocol and ActivityPub
  • Current limitations: chat/DMs centralized; most users still on Bluesky's infrastructure; full relay requires 16TB NVMe

Source: Backlinko, Sprout Social, Business of Apps, Nieman Lab

Reddit (Post-IPO)

IPO: March 2024

Full Year 2025 Results (reported February 5, 2026):

  • Total revenue: $2.2 billion (+69% YoY)
  • Q4 revenue: $726M (+70% YoY)
  • Ad revenue: $2.1B full year (+74% YoY); $690M in Q4 (+75% YoY)
  • Net income: $530M full year (24% margin); $252M in Q4 (+255% YoY)
  • Adjusted EBITDA: $845M (38% margin)
  • Free cash flow: $684M (3x 2024)
  • Diluted EPS: $2.62 (vs. loss in prior year)
  • DAU: 121M+ in Q4 (+19% YoY)
  • Weekly active users: 471M+ (+24% YoY)
  • Monthly users: estimated 1.1-1.36 billion
  • Premium subscribers: 6.8M (+92% in two years)
  • Announced $1B share repurchase program
  • Q1 2026 guidance: $595-605M revenue (52-54% growth)
  • Cash on hand: $2.5B

Data Licensing / Enshittification:

  • $60M/year Google deal to license data for AI training
  • IPO prospectus disclosed $203M in data licensing arrangements (2-3 year terms, signed January 2024)
  • API pricing: $0.24 per 1,000 requests for commercial use (free tier extremely limited); enterprise pricing opaque, requires sales contact
  • 2023 API pricing changes killed most third-party apps and independent tools
  • Revenue-sharing pilot for top creators; potential paywalls for exclusive subreddits under exploration

Source: Reddit Q4 2025 Earnings, CNBC, SociallyIn, TechCrunch

Snapchat

Full Year 2025 Results (reported February 4, 2026):

  • Revenue: $5.93B (+11% YoY)
  • Q4 revenue: $1.72B (+10% YoY)
  • Q4 net income: $45M; full year net loss: $460M (improved from $698M loss in 2024)
  • Free cash flow: $437M (2x 2024)
  • MAU: 946M (Q4 2025, +6% YoY) — approaching 1 billion
  • DAU: 474M (Q4 2025, but lost 3M vs. Q3 due to Russia ban and Australia under-16 restrictions)
  • Snapchat+ subscribers: 24M (+71% YoY), generating ~$750M/year
  • All user growth from "Rest of World" — US and Europe appear saturated
  • AR Specs consumer launch planned for 2026
  • Q1 2026 guidance: $1.50-1.53B (below analyst estimates)

Source: Snap Investor Relations, CNBC, Music Ally

LinkedIn

  • 1.2B+ registered members (2026, +9% YoY); ~310M active users
  • 175M Premium users (up from 154M in 2022)
  • US leads with 239M profiles; India second with 155M
  • Crossed $5B quarterly revenue in Q4 2025 (~$20B+ annual run rate)
  • 2025-2026: enhanced video content (+34% upload growth YoY), AI recruitment tools, verified identity badges, Calendly integration, AI-powered Predictive Career Path Analysis tool
  • January 2026 algorithm update: shifted to signal weighting (dwell time, "See more" expansion rate, comment quality over volume)
  • 2024 algorithm change reduced reach for 95% of creators by ~50%
  • 18,000+ employees, 69M+ companies listed, 24,000+ LinkedIn Learning courses

Source: Cognism, Wave Connect, DemandSage, HeyOrca

Mastodon / Fediverse

  • Mastodon: ~1-1.5M monthly active users; ~9-12M registered accounts across the fediverse (excluding Threads)
  • Growth has plateaued — active users in early 2025 reportedly lower than two years prior, despite brief influx around Trump inauguration
  • 27% of users based in Germany; 66% male; 25-34 age group largest at 30%
  • Mastodon working on account renaming (numeric IDs), Quote Posts, and Fediscovery (search/discovery service)
  • Threads introduced ActivityPub crossposting in 2024 (outside EEA); full two-way compatibility still incomplete
  • WordPress ActivityPub plugin expanded significantly in 2025 (Following feature, moderation tools)
  • Criticized as "unapproachable for new users" and having "changed very little in recent years"
  • Bluesky's AT Protocol growth is seen as competitive pressure; some argue fediverse needs plurality beyond just ActivityPub

Source: FediDB, ThinkImpact, Manton Reece, Diverse Tech Geek


The Enshittification Timeline

  • 2006-2011: The golden era. Facebook, Twitter, Instagram launch with chronological feeds. Users see what they follow. The product serves the user.

  • 2012: Facebook goes public. Algorithmic feed replaces chronological. Organic reach for pages begins declining — from ~16% to under 2% by 2016. The message to businesses: pay to reach people who already chose to follow you.

  • 2014-2018: Instagram shifts from chronological to algorithmic feed. Stories copied from Snapchat. Explore tab pushes content from strangers over friends. The product begins serving the algorithm, not the user.

  • 2018: Cambridge Analytica scandal reveals Facebook shared data of 87 million users with a political consulting firm. The data practices that enabled this were features, not bugs.

  • 2020-2023: TikTok forces all platforms to prioritize short-form video and algorithmic recommendations over social graphs. Your feed becomes content from people you don't follow. "Social" media becomes "media" — the social part is incidental.

  • 2023: Twitter/X under new ownership removes legacy verification, restricts API access (from free to $42,000/month for basic access), kills third-party apps, and modifies content moderation. Trust decline accelerates. Reddit kills its free API tier, destroying most third-party apps overnight. Reddit discloses $203M in data licensing deals — selling user-generated content to AI companies.

  • 2024: Meta introduces AI-generated content in feeds. Instagram pushes "Recommended" content to 50%+ of feed. Users report seeing less from friends, more from brands and algorithmic suggestions. Facebook begins showing AI-generated comments. Reddit IPOs (March 2024) and begins optimizing for Wall Street. X/Twitter defaults all users into Grok AI training on their posts — no notification, opt-out buried in settings. LinkedIn algorithm change reduces reach for 95% of creators by ~50%.

  • 2025 (actual events):

    • January 7: Meta announces end of third-party fact-checking, replaced by X-style Community Notes. Mark Zuckerberg cites "cultural tipping point" after Trump's election. Relaxes content policies on immigration and gender topics. Moves US content moderation team from California to Texas. Ends internal DEI programs. (CNN, Meta blog)
    • January 17: Supreme Court upholds TikTok divest-or-ban law. TikTok briefly goes dark January 18-19 before Trump executive order restores service.
    • May-December: Grok (X's AI chatbot) used to generate 6,700 sexually suggestive or nonconsensual deepfake images per hour — 84x more than the top 5 deepfake websites combined. Indonesia, Malaysia, Philippines temporarily block Grok. UK PM Starmer says banning X is "on the table." (The Conversation)
    • June 27: Supreme Court upholds Texas age-verification law for adult content (Free Speech Coalition v. Paxton), opening door for broader social media age verification.
    • AI slop at scale: Usage of the term "AI slop" increases 9x YoY. Named "Word of the Year" by Macquarie Dictionary, Merriam-Webster, and American Dialect Society. Facebook hit hardest — older users (24% of base is 55+) most vulnerable to AI-generated scams. Pinterest introduces tools to limit AI content in feeds. (Meltwater)
    • July: AI-generated band "The Velvet Sundown" (1M+ Spotify listeners) revealed as entirely synthetic. Guess runs AI-generated model ad in Vogue, sparking backlash. McDonald's Netherlands pulls AI-generated TV commercial after 3 days.
    • December: EU issues first major DSA fine: 120M on X for deceptive "blue checkmark" design. US Secretary of State sanctions former EU commissioner who drafted the DSA. Meta updates privacy policy (effective Dec 16) to use AI chat interactions for ad targeting. (Taylor Wessing)
    • December 18: TikTok divestiture deal signed with US investor consortium.
    • Reddit posts $2.2B revenue (+69% YoY), $530M net income — first full profitable year post-IPO.
  • 2026 (events through March):

    • January 22: TikTok USDS Joint Venture officially established. ByteDance retains 19.9%. TikTok immediately updates terms to expand data collection.
    • February 3: French prosecutors raid X's Paris offices over Grok deepfake crisis.
    • February 5: Trump shares racist AI-generated video of the Obamas on X; deleted next morning after bipartisan backlash. (Crescendo AI)
    • EU DSA enforcement intensifying: 14 investigations opened; platforms now conducting compliance workshops behind closed doors; Commission describes relationship with platforms as "more adversarial."
    • Platforms push AI agents that interact on users' behalf — posting, commenting, responding to messages. The "social" in social media increasingly means algorithms talking to algorithms, with humans as the audience.
    • Meta projects $115-135B in 2026 capital expenditure for AI infrastructure. Zuckerberg declares goal of "personal superintelligence."

The Data Audit

Meta: Privacy Policy and AI Training (2025-2026)

What Meta collects (representative of the category):

  • Location history (GPS, Wi-Fi, cell towers)
  • Browsing behavior across the web (via Meta Pixel, installed on millions of sites)
  • Purchase data (from partners and payment integrations)
  • Biometric data (facial recognition, voice patterns)
  • Contact lists (uploaded from phone)
  • Message metadata (who you talk to, when, how often — even in "encrypted" Messenger)
  • Device information (battery level, storage, installed apps, hardware identifiers)
  • Cross-app tracking (across Facebook, Instagram, WhatsApp, Threads, and third-party apps)
  • NEW (December 16, 2025): AI chat interactions across Facebook, Instagram, Messenger, and WhatsApp used to personalize recommendations and ads — the first time a major platform leverages active AI conversations for ad targeting at scale

Meta's AI training policy (current):

  • All public posts, photos, comments, and reels can be used to train Meta's AI models
  • Meta states: "We do not use the content of your private messages with friends and family to train our AIs unless you or someone in the chat chooses to share those messages with our AIs."
  • However, privacy advocacy group noyb characterized Meta's policy as stating the company "wants to take all public and non-public user data that it has collected since 2007 and use it for any undefined type of current and future 'artificial intelligence technology.'"
  • Meta AI may use information about non-users if their information appears in other users' public posts
  • Users cannot fully opt out; certain sensitive topics (religion, health, politics) excluded from targeting
  • David Evan Harris (UC Berkeley AI ethics): "Because the US has no federal regulations about privacy and AI training, people have no standardized legal right to opt out of AI training."
  • Due to GDPR, the AI data usage rules do not immediately apply in the EU, UK, or South Korea. Irish DPC monitors rollout of objection forms.

Source: Snopes, Social Media Today, Visualping, Al Jazeera, noyb

How it's monetized: This data powers the most precise advertising targeting system ever built. Advertisers can target you by: relationship status, political views, income level, health interests, purchasing behavior, life events (new job, new baby, recent move), and thousands of other dimensions. You cannot meaningfully opt out while using the service — the data collection IS the service, from Meta's perspective. Meta's advertising revenue was $200B+ in 2025, representing ~97% of total revenue.

What happens if acquired or policies change: Meta's data — representing billions of intimate human relationships, private conversations, and behavioral patterns — transfers to whoever controls the company. There is no user consent mechanism for this transfer. The data exists. Whoever owns the servers owns the data.

X/Twitter: Data Practices Under Musk

  • X trains Grok AI on 500 million daily tweets — including deleted ones — without explicit user consent
  • Data sharing for AI training opted in by default; 73% of X users were unaware their tweets trained Grok
  • Opt-out mechanism exists but is "technically present but practically invisible" (buried in Settings > Privacy and Safety > Grok)
  • Updated Terms of Service (November 15, 2024): explicitly grants X the right to use user data for AI training once users agree
  • Privacy Policy updated to allow third-party "collaborators" to train AI models on X data unless users opt out — X is looking into licensing data to AI companies (following Reddit's model)
  • 2025 policy pivot: banned external scraping of public data, but reserves right to use all content internally for Grok
  • EU: Irish DPC found X "unlawfully using personal data of more than 60 million users in EU/EEA to train AI without consent"; X agreed to permanently stop processing European users' data for Grok

Source: Cambridge Analytica (blog), Silicon Republic, TechCrunch, Malwarebytes

TikTok: Data Practices and Government Findings

  • FTC investigation into whether TikTok violated federal law prohibiting "unfair and deceptive" practices by denying Chinese access to US user data
  • Reports of keystroke monitoring (typing pattern analysis)
  • Supreme Court found ByteDance "subject to Chinese laws that require it to assist or cooperate with the Chinese government's intelligence work" and that TikTok's data collection from 170M+ US users "could be exploited for surveillance, public influence campaigns, or other harmful purposes"
  • Irish DPC fined TikTok 345M (2023) for mishandling children's data
  • UK ICO fined TikTok 12.7M for processing data of children under 13
  • Post-deal (January 22, 2026): TikTok updated terms expanding demographic data collection and precise location tracking under US ownership
  • EU Digital Services Act now requires TikTok to disclose algorithm operations and mitigate misinformation risks

Source: PBS, Holland & Knight, Human Rights Watch, Washtenaw Voice

Reddit: Data Licensing and API Monetization

  • $60M/year Google deal for AI training data
  • $203M in data licensing contracts disclosed at IPO (2-3 year terms)
  • API pricing: $0.24 per 1,000 requests (commercial); free tier "extremely limited"
  • Enterprise pricing opaque — requires contacting sales team
  • Pricing "designed to extract maximum value from large commercial operations rather than supporting the diverse ecosystem of smaller developers, researchers, and hobbyists"
  • Third-party apps effectively killed by 2023 pricing changes
  • Revenue-sharing pilot for top creators; potential paywalls for exclusive subreddits under exploration

Source: TechCrunch, AutoGPT, Sahm Capital

Recent incidents (all platforms):

  • 2021: Facebook data leak exposes personal data of 533 million users
  • 2023: Meta fined $1.3B by EU for GDPR violations (data transfers to US)
  • 2023: TikTok fined 345M by Irish DPC for children's data violations
  • 2024: X accused of unlawfully using 60M+ EU users' data for Grok AI training
  • 2024-2025: Multiple lawsuits regarding use of user content for AI model training without explicit consent
  • 2025: Meta updates privacy policy to use AI chat data for ad targeting
  • 2025-2026: Grok deepfake crisis — nonconsensual sexually explicit AI images at industrial scale
  • 2026: TikTok expands data collection terms immediately after US ownership transfer

Social Media Regulation (2025-2026)

EU Digital Services Act (DSA) Enforcement

  • 14 formal investigations opened by the European Commission as of November 2025 (platforms include AliExpress, Facebook, Instagram, Temu, TikTok, X)
  • First major fine: 120M on X (December 2025) for deceptive "blue checkmark" verification design and transparency violations
  • EU piloting age verification app (Denmark, France, Greece, Italy, Spain) with broader rollout expected 2026
  • Digital Omnibus proposal (November 2025) moves AI oversight for large platforms to the Commission's AI Office as "one-stop-shop"
  • Digital Fairness Act delayed to Q4 2026

Source: MediaLaws, Taylor Wessing, King & Spalding

US-EU Tensions

  • December 2025: US Secretary of State Marco Rubio sanctioned former EU commissioner Thierry Breton (who led drafting of the DSA) — froze assets, designated persona non grata
  • July 2025: House Judiciary Committee report accusing EU of "weaponizing" DSA for "global censorship"
  • Three former EU commissioners issued joint response (January 2, 2026) defending DSA and DMA as anti-monopoly measures, not speech policing
  • DSA enforcement workshops now held behind closed doors at company request; Commission describes work as "more adversarial" and staff using disappearing messages on Signal

Source: Politico, PPC Land, House Judiciary Committee

US Legislation

  • Kids Online Safety Act (KOSA): Reintroduced in both House and Senate (2025 versions). Requires platforms to implement safeguards for "known" minors. Does not explicitly require age verification. Eliminates prior "duty of care" language. Still debated over constitutional concerns.
  • COPPA 2.0 (S. 836): Reintroduced March 2025 by Senators Markey and Cassidy (20 cosponsors). Raises coverage from under-13 to under-17. Prohibits targeted advertising to minors. Requires "eraser button" for minors' data. Passed Senate Commerce Committee (June 2025, voice vote).
  • FTC COPPA Rule update: Finalized January 16, 2025. Effective June 23, 2025; full compliance deadline April 22, 2026. Now covers biometric identifiers (fingerprints, facial data, voice data) and government-issued identifiers.
  • Supreme Court (Free Speech Coalition v. Paxton, June 27, 2025): Upheld Texas age verification law for adult content (6-3). Opens door for broader age verification mandates.
  • State-level activity: 25 states have passed age verification laws for porn sites. Utah enacted first app store age verification law. California and Minnesota enacted social media warning label laws (effective 2026-2027).
  • Kids Off Social Media Act (KOSMA) and Stop the Scroll Act pending in Congress.
  • No comprehensive federal privacy law. Section 230 protections remain in tension with state content moderation laws (Texas, Florida).

Source: Davis Wright Tremaine, Mayer Brown, Inside Privacy, ACLU

EU AI Act

  • Comprehensive risk-based framework with transparency mandates for generative AI
  • Requires disclosure of AI-generated content and copyright compliance
  • Most provisions enforceable starting 2026
  • South Korea's AI Basic Act entered force January 2026

AI-Generated Content Controversies (2025-2026)

The Grok Deepfake Crisis

The defining AI-social media scandal of 2025-2026. Elon Musk's Grok chatbot on X was used to nonconsensually alter images — including of minors — to depict them in sexually suggestive contexts. By January 2026, users were generating 6,700 sexually suggestive or nudified images per hour through Grok (84x more than the top 5 deepfake websites combined). Government responses: Indonesia, Malaysia, Philippines temporarily blocked Grok; UK PM Starmer stated banning X was "on the table"; French prosecutors raided X's Paris offices (February 3, 2026).

Political Deepfakes

  • February 5, 2026: Trump shared racist AI-generated video of Barack and Michelle Obama on X; deleted next morning after bipartisan backlash
  • July 2025: Trump reposted AI deepfake of Obama being arrested
  • January 2026: After Venezuelan President Maduro's capture, misleading AI content generated millions of views

"AI Slop" at Industrial Scale

  • Term usage up 9x in 2025 vs. 2024
  • Named Word of the Year by Macquarie Dictionary, Merriam-Webster, and American Dialect Society (December 2025)
  • Facebook hardest hit — older users (24% of base is 55+) most vulnerable to AI scams
  • Pinterest introduced tools (October 2025) to let users limit AI content in feeds
  • Deepfake videos estimated to surge from ~500,000 (2023) to 8 million (2025) — 16x increase (European Parliamentary Research Service)
  • NewsGuard: leading AI chatbots spread false information 35% of the time on controversial topics

Corporate/Cultural Backlash

  • "The Velvet Sundown" — AI-generated rock band with 1M+ Spotify listeners — exposed July 2025
  • Guess AI-generated model advertisement in Vogue (August 2025) — backlash over human displacement
  • McDonald's Netherlands AI commercial pulled after 3 days (December 2025)
  • Fan art communities organizing resistance against AI integration on platforms

Source: Crescendo AI, Stimson Center, PBS, Meltwater, Digital Watch Observatory, The Conversation


Vulnerability Score

CriterionRatingExplanation
User resentmentHighSurvey after survey shows declining trust and satisfaction. Users stay because of network effects, not preference.
Switching costHighYour friends are there. Your photo history is there. Rebuilding a social graph is the highest switching cost in technology.
Technical feasibilityMediumThe software is buildable. The social graph is not. A competitive product needs a community-first strategy, not a feature-first strategy.
Monetization clarityMediumUsers don't pay for social media today. A subscription model ($3-5/month) is possible but unproven at scale. Ad-free, user-owned social requires users to accept paying.
Data sensitivityVery HighSocial media data is among the most intimate: relationships, beliefs, private moments, location patterns, communication habits.
Network effectsVery HighThe strongest of any category. The product is the people on it.

Overall vulnerability: Medium-High. User resentment is extreme and growing. But network effects create the strongest moat in technology. The path is not to compete head-on with Instagram or TikTok — it's to build for the communities that are actively seeking alternatives: privacy-conscious users, creators tired of algorithmic suppression, communities around shared interests who want to own their space.

A Your 99 social platform doesn't need to replace Instagram. It needs to be the place where 100,000 people who care about owning their social space choose to connect.


The Your 99 Blueprint

Revenue model: Optional premium subscriptions ($3-5/month for additional features like extended storage, analytics for creators, custom communities). No advertising. The product is funded by users, not by selling users.

Draft Contribution Map:

ContributionStake per month
Active use (10+ days/month)10 base units
Content creation (engaged with by others)5-50 units (scaled by genuine engagement)
Community moderation (verified)20 bonus units
Harmful content reporting (verified)5 bonus units
Premium subscription30 base units
Referral (becomes active 30+ day user)15 bonus units

Economics at scale:

ScaleUsersPremium %Monthly RevenueDistributablePer User (avg)
Small10,00020%$8,000$5,700$0.57
Medium100,00020%$80,000$57,000$0.57
Large500,00020%$400,000$285,000$0.57

(Assumes $4/month average premium, 10% operating costs, standard 1%/10%/89% split)

Per-user returns are modest for social media alone. The real value proposition is: ad-free, algorithm-transparent, user-governed, data-sovereign social media — plus ecosystem-wide earnings from the Universal Pool across all Your 99 products.

Key differentiator beyond ownership: Chronological feeds by default. Algorithmic discovery opt-in, not opt-out. Content moderation governed by the community, not by a content policy team in Menlo Park. Your data exportable at any time. No AI training on your content without explicit governance approval.

Minimum viable feature set: Profiles, posts (text + images), follows, chronological feed, community groups, basic search. No stories. No reels. No algorithmic feed. Simplicity is the feature.


Open Questions

  • How do you compete with algorithmic content recommendation without the data moats incumbents have? Is "no algorithm" itself the selling point for the first users?
  • Does a user-owned social platform need content moderation governance, and how does that interact with the Your 99 governance model? Where is the line between community standards and censorship?
  • Can you bootstrap a social network without importing the social graph? Or do you need to build around communities (interest-based) rather than social connections (friend-based)?
  • Is the subscription model viable for social media, or do users expect social to be free? Would a freemium model with ownership for all users work?
  • Should this be an early Your 99 product (high visibility, strong narrative) or a later one (after proving the model with lower-network-effect categories)?

Report version 0.2

Last updated 2026-03-01