The Anti-AI Social Media Platform of our Dreams is here!

Overview

The video is a promotional Q&A/stream segment introducing Haven Social, a planned anti-AI and privacy-focused social media platform. Haven is positioned as an “ethical antidote” to mainstream platforms such as Instagram, TikTok, Twitter/X, and YouTube.

The core claim is that social media has become more exploitative over time—particularly through:

• Ads
• Surveillance, including facial recognition
• Use of user content to train AI systems

Haven’s pitch is that it will reverse this trend by design.

Haven’s Main Differentiators (Per the Founder)

Anti-surveillance / anti-facial recognition

Haven claims it will protect users from facial recognition and mass surveillance, including threats exemplified by Clearview AI (described as identifying people across the internet using photos).

No AI training on user content

The platform says it prevents users’ art and media from being used to train AI models. It also frames itself as avoiding “shoving AI features” onto users—contrasted with YouTube-style AI filters.

No covert algorithm manipulation

Haven emphasizes transparency, claiming it won’t silently manipulate feeds “behind the scenes.”

No generative AI allowed

Haven repeatedly states that generative AI content will be disallowed, presenting this as essential for trust and identity in an anti-AI platform.

Anti-AI “Poisoning” for Artist Protection

A major section focuses on protecting artists from AI scraping and training using data poisoning (tools mentioned: Mist, Glaze, Nightshade).

The pitch includes:

• Artists often can’t use poisoning tools widely because they require significant GPU resources.
• Haven will provide a free, built-in poisoning pipeline: users upload art, and Haven applies poisoning automatically.
• The founder argues poisoning can be effective at relatively small scale, claiming AI models can be disrupted by limited poisoned samples (example claim: “200 poison samples” affecting very large datasets).

Detecting/Removing AI “Slop” and Moderation Approach

Haven claims it will address “AI slop” via multiple layers:

• Technical detection
  • Including removal of watermark-like signatures such as “SynthID” (described as being embedded in some AI outputs).
• Manual/report-based moderation
  • Because AI detectors are described as “hit or miss.”
• User incentive system
  • Users answer prompts about whether content violates rules.
  • They can earn a small reward and donate it to creators.
  • The goal is to improve detection and incentivize reporting.
• Algorithmic moderation via enforcement incentives
  • The founder argues AI spam persists elsewhere because platforms don’t reliably remove/penalize it, especially when revenue rewards engagement.

Governance, Funding, and “Core Values”

Haven argues its rules and values won’t be “overwritten” because it plans to be financially independent from VC control.

Funding plan (as described):

• Advertising
  • Kickstarter is used for initial scaling
  • Ads fund ongoing operations and R&D

Ethical ad handling claims include:

• No selling user data externally (keeping it internal/private)
• Ad relevance is justified as necessary for funding
• A paid/subscription mode is described as reducing data collection (“turn off any data collection”)

Content Policy and Censorship Stance

Haven calls itself anti-censorship, but clarifies it does not mean “absolute free speech” (i.e., not allowing slurs or death threats).

The framing is that “anti-censorship” means resisting:

• Shadow suppression
• Algorithmic suppression by platforms/governments

Other policy points:

• NSFW/nudity is largely accepted as part of art
  • Moderation is described as audience-targeting based on preferences
• Gore is allowed in context (e.g., educational/medical or artistic depictions)
  • Not allowed as “real-world execution compilations”
• Spoiler/content warnings and blurring are described as features for sensitive material

Scraping and Platform Abuse Prevention

The founder claims scraping can be deterred using ideas like:

• Cloudflare’s AI “labyrinth”
  • Described as trapping scrapers in endless non-valuable loops

They also argue that simply banning AI isn’t enough: platforms that ban AI but provide no real defenses leave themselves attractive to scrapers.

Anti-Competitor Skepticism (Claims and Allegations)

The founder expresses skepticism about other “anti-AI” platforms and raises concerns about inconsistencies.

He discusses a competitor (referred to in subtitles as “Devine”/“Shakespeare” and related company details as presented in the subtitles) that is described as:

• Offering anti-AI content restrictions
• Also providing AI-related tools

The segment implies suspicion that the same organization may be involved in AI training, including claims about the value of human-only data.

These points are presented as doubts about whether such competitors are truly aligned with an anti-AI mission.

Growth and Overcoming the “Network Effect”

To attract non-art users and avoid being “artists-only,” Haven proposes:

• Targeting privacy/security-focused users
• Reaching people broadly tired of mainstream platform “enshitification”

To reduce migration friction, Haven suggests:

• Importing content and follower relationships from other platforms:
  • Verify ownership of an existing account
  • Automatically import posts (with timestamps)
  • Optionally migrate follower lists so communities don’t need to rebuild from scratch

Monetization

Haven’s monetization model (as described) is similar to YouTube:

• Once creators reach a threshold, they can apply for monetization based on views.
• The founder emphasizes that, unlike some other platforms (examples claimed), Haven will have ad revenue to fund creator payouts.

Presenters / Contributors

• William Lavel — Founder of Haven Social
• Lucky Inkyio — Co-host
• Video host/narrator — Interviewing/introducing (name not provided in the subtitles)