8 myths about AI and social media, checked against the sources

Most worries about AI on social media come from second-hand rumours. Here are the eight we hear most, checked against the platforms' own pages.

1. "Platforms shadowban AI-written captions"

False. None of Instagram, Facebook, Threads, TikTok, YouTube, LinkedIn or X says AI-written text is ranked lower. They demote copied, reposted and spammy content (the full breakdown).

2. "Labelling AI content kills your reach"

False for YouTube, which says so directly. "Disclosing AI content alone won't limit a video's audience" (YouTube).

3. "You must label every AI-written post"

False. Labels cover realistic AI images, video and audio, not captions or replies (AI labels by platform).

4. "Any automated DM breaks Instagram's rules"

False. Meta's official API allows automated replies within 24 hours of a message, and one private reply per comment within 7 days (Meta). Mass DMs to strangers and password-based bots are what break the rules (is automation allowed?).

5. "Replying to comments doesn't matter"

False. Comments and replies appear in the ranking explainers of Instagram, Facebook, Threads, TikTok and X (do replies boost reach?).

6. "Google penalises AI-written blog posts"

False. Google judges content quality, not how it was produced. Using AI mainly to manipulate rankings is what breaks its spam policies (Google).

7. "Customers hate AI replies"

It depends on the reply. 64% would prefer companies didn't use AI for service, mostly from fear of not reaching a person (Gartner, 2024). Yet 69% are comfortable with AI that delivers faster care (Sprout, 2025, vendor data). Accurate answers with a path to a human win both groups (details).

8. "AI means you can post anything, as much as you like"

False. Facebook deprioritises duplicate content, and YouTube's inauthentic content policy targets mass-produced, template-made AI videos (Meta, 2026, YouTube). AI saves time; it doesn't replace having something to say.

The rule behind all eight

Platforms reward what people want and punish shortcuts that fake it. Use AI to answer faster and post more consistently, about real things, and the rules are on your side.