Does AI content get penalised? What the platforms actually punish

In threads about slumping engagement you'll now reliably find the claim, stated as fact, that "Instagram and Facebook suppress AI-written posts and AI images". Industry posts circulate figures like thirty per cent less reach for "low-effort AI" content on LinkedIn. Is any of it true? The honest answer needs the folklore separated from the documentation.

What's documented

Platforms document quality signals, not authorship detection. Meta, LinkedIn and TikTok all describe downranking low-quality, unoriginal, engagement-bait content — categories that generated sludge falls into by default but is not the definition of. Platforms also label AI-generated media (increasingly via embedded credentials) — labelling is disclosure, and no platform documents a blanket ranking penalty for the label itself. Meanwhile ranking has shifted toward dwell time and genuine interaction, which punishes generic content regardless of who wrote it.

Why the folklore feels true

Accounts that adopt AI badly change their output overnight: more posts, blander posts, samer posts. Reach drops. The tool gets the blame, but what the algorithm saw was a quality drop and an audience that stopped dwelling. A human posting the same generic volume would slide identically — it just used to be too expensive to test that.

Diagnose with benchmarks, not anecdotes

Before accepting "we're being suppressed", compare engagement rate against your industry's typical range (organic ER by platform is well published; it's built into Brandaris analytics for exactly this reason). Below benchmark across all content: strategy problem. Fine on some posts and dire on the generic ones: content problem — and a fixable one.

The recovery sequence

Cut the volume. Ground every post in something specific — a real conversation from your listening, a real product fact, a real photo. Put an editor's judgement back between draft and feed. Accounts recover on the same signals they slid on. The platforms aren't punishing your tool; they're pricing your genericism.