Socialync Logo
Socialync

AI Features & Disclosure

Last Updated: April 30, 2026

Socialync uses AI to help you draft, refine, and repurpose content. This page explains, in plain language, which features use AI, where your data goes, what your responsibilities are, and the rules we enforce so AI never publishes on your behalf without your say-so.

For the binding versions, see the Terms of Service (Section 8) and the Privacy Policy (Section 5).

1. Which features use AI

  • Caption and post drafting — turn an idea, prompt, or topic into a draft caption tailored to a platform's tone and length
  • Content suggestions and rewrites — improve, shorten, lengthen, or repurpose drafts you've started
  • Style analysis — analyze your existing content to learn your brand voice and reuse it on new drafts
  • Transcript analysis — extract clips, themes, and post ideas from a transcript or video
  • MCP integrations — connect external AI agents (e.g. Claude desktop, custom agents) that draft posts into your Socialync drafts queue

2. The big rule: humans publish, AI drafts

If you're posting or scheduling content yourself, nothing changes. Manual posting and manual scheduling work the way they always have. Write the post, hit publish or schedule, done.

If an AI is directing the post — drafting the copy, generating the image, or telling Socialync to publish or schedule on your behalf — a human in your account has to review and approve it before it goes anywhere. This applies whether the AI is trying to publish immediately or just trying to put posts into your schedule.

This is enforced by the product, not just the policy:

  • AI features and MCP agents create drafts, never live or scheduled posts
  • Drafts land in the drafts approval queue for a human to review
  • Only after a human approves a draft can it be scheduled or published
  • Building a script or bot that auto-clicks "approve" violates the Terms of Service

3. What data goes where

3.1 AI sub-processors

We use the following AI providers:

  • Anthropic (Claude) — drafting, content generation, MCP-based interactions
  • OpenAI (GPT) — drafting, content generation, transcription
  • Other providers may be used for specific features (e.g., transcription, image generation); we'll name them in the relevant feature documentation

3.2 What gets sent

  • The content you submit to the feature (your prompt, draft, transcript, etc.)
  • Style or brand preferences you've configured, where relevant to the feature
  • Minimal operational metadata to route the request

3.3 What does NOT get sent

  • Your platform OAuth tokens or social media account credentials
  • Your payment information
  • Your followers' personal data

3.4 No training on your content

We do not use your content to train AI models. We use AI sub-processors under their commercial API terms, which by default do not train on data submitted through the API. Each provider's published API terms govern the specifics.

3.5 Retention

  • AI providers may retain inputs and outputs short-term (typically up to 30 days) for abuse monitoring
  • Drafts saved in your Socialync account stay until you delete them
  • We don't keep a long-term archive of AI inputs and outputs

4. User-connected AI agents (MCP)

If you connect a third-party AI agent or MCP-compatible service to your Socialync account (e.g., Claude desktop with our MCP server), that agent is operated by you. Data the agent processes is governed by the terms of whatever AI service you've connected.

What this means in practice:

  • You're the controller of data flowing through your own AI agent
  • You're responsible for the agent's behavior, including content it drafts and any disclosure obligations
  • Your agent still goes through our human-approval queue — it can draft, but it can't publish
  • Posts published through an MCP-connected agent are treated as user-initiated and user-authored under our Terms

5. Disclosure: when to label content as AI-generated

If you're publishing AI-generated or AI-modified content, you may have to disclose that fact. Disclosure obligations come from two places:

5.1 Destination platforms

Each connected platform has its own AI-content disclosure rules. As of writing:

  • TikTok requires labeling realistic AI-generated content depicting people, places, or events
  • Meta (Facebook / Instagram) requires labeling photorealistic AI-generated images, video, and audio of real people or events
  • YouTube requires creators to disclose meaningfully altered or synthetic realistic content during upload
  • X, LinkedIn, Threads, Bluesky have their own evolving policies — check the platform documentation when in doubt

Where Socialync surfaces a platform's AI-disclosure flag at posting time, please use it. Where the destination platform does not yet expose a disclosure flag through its API, you may need to disclose in the post copy itself.

5.2 Applicable law

  • EU AI Act — if your audience includes EU users, providers and deployers of generative AI must label synthetic content (e.g., images, audio, video, text) as AI-generated where it could be mistaken for authentic. Deepfakes have specific labeling rules.
  • State and national laws — some jurisdictions have additional rules about AI in political ads, AI in health/financial content, and synthetic media depicting real people. You're responsible for complying with the laws that apply to you and your audience.

Socialync provides tooling to help you comply, but does not guarantee compliance on your behalf.

6. What you may not do with AI on Socialync

  • Publish AI-generated content without prior human review and approval
  • Bypass, disable, or script around the drafts-approval step
  • Generate non-consensual intimate imagery (NCII), including AI-generated
  • Generate deepfakes of real, identifiable people without their documented consent
  • Use AI to impersonate real people, brands, or entities
  • Use AI to falsely depict public figures in criminal, sexual, violent, or otherwise harmful contexts
  • Generate content that sexualizes minors in any form
  • Use AI to run spam, scams, fraud, or coordinated inauthentic behavior

See the Acceptable Use Policy for the full list and rationale.

7. AI accuracy and your responsibilities

AI outputs can be wrong. They can hallucinate facts, misattribute quotes, invent statistics, or produce content that looks fine but isn't. When you use Socialync's AI features:

  • You remain the author and publisher of any post you ship
  • You're responsible for fact-checking AI-drafted content before approving it
  • You're responsible for editing or rejecting outputs that are wrong, off-brand, or otherwise unsuitable
  • You're responsible for any disclosures the destination platform or applicable law requires

8. Reporting AI-related abuse

If you see content distributed through Socialync that appears to be a deepfake, NCII, AI-generated impersonation, or another AI-driven abuse, report it to safety@socialync.io. Valid reports of non-consensual intimate imagery are actioned within 48 hours, in line with the US Take It Down Act.

9. Related documents