The wave of social platforms cracking down on AI-generated content accelerated on 31 July, when Snap announced it would stop recommending or rewarding fully AI-generated videos in Spotlight, its short-form feed, shifting its algorithm to prioritise content made by real people.
Social Platforms Cracking Down: What the Policies Actually Say
Snap’s move is part of a broader tightening across the creator economy. YouTube, Substack, TikTok, Pinterest, Meta, and LinkedIn have each updated their approach to machine-generated content in recent months, ranging from labelling requirements to outright demonetisation.
YouTube’s changes are more precise than a simple ban. YouTube’s support pages confirm the platform is formally renaming its ‘repetitious content’ policy to ‘inauthentic content’ under its YouTube Partner Programme guidelines, while making clear there are no changes to its existing ‘reused content’ rules. Commentary, clips, compilations, and reaction videos can still monetise, provided the creator has added significant original value.
A YouTube spokesperson told Business Insider the policy is not new but an important clarification of its long-standing rules, intended to reduce ambiguity for creators who have been demonetised for repetitive, low-effort, or emotionally manipulative videos. The platform had already removed more than a dozen channels that amassed millions of views by pumping out AI-generated content featuring characters like cats and Jesus. Video editing platform Kapwing estimated that 21% of YouTube’s feed consisted of AI-generated videos.
Substack took a different route. On 21 July, the newsletter platform said it would partner with Pangram, an AI detector tool, to let readers scan content and estimate how much AI may have been used in its writing. Creators can add a disclosure explaining their process, such as noting they used Claude for copyediting, or flag that they use no AI at all.
Pangram’s track record, however, is disputed. Writer Tim Requarth has documented cases where the detector was at the centre of high-profile controversies, including flagging the ‘Shy Girl’ novel that Hachette subsequently cancelled and flagging a Modern Love column that went viral. Separately, an AI Weekly newsletter citing research from Epoch AI found that AI detectors, including Pangram, missed up to 18% of AI-generated text, underlining that detection remains probabilistic rather than definitive.
There is also a timing wrinkle. An independent analysis found that Substack’s Privacy Policy was updated on 14 May 2026, quietly adding ‘generative AI services’ to its list of third-party service provider categories, with no announcement and no click-through consent required from writers. That update preceded the Pangram feature’s public launch on 21 July by more than two months. Creators can opt out of Pangram’s automated scanning in their publication settings, which also removes the AI-use estimate from their posts for readers.
Snap, LinkedIn, and the Creator Economy Squeeze
Snap’s Spotlight has come a long way since its debut. When the platform originally launched Spotlight in November 2020, Snap’s investor newsroom confirmed the company committed to distributing over $1 million USD every day to Snapchatters who created the top Snaps on Spotlight, at least through the end of that year. The current policy shift moves the platform firmly in the other direction: fully AI-generated videos will no longer qualify for recommendations.
‘Spotlight should remain a place where people can discover authentic creativity from real people,’ Snap said in a blog post. Creators can still use Snap’s AI tools to enhance or edit videos, but fully AI-generated content is excluded from recommendation eligibility.
LinkedIn presents a starker picture. AI detector Pangram estimated that 41% of long-form LinkedIn posts and 30% of short-form posts from April through June were AI-generated, the highest share among platforms it analysed, including X, Reddit, Substack, and Medium. The platform actively promotes AI writing tools, including a button that helps users polish posts with AI, yet said in a June post it would be taking steps against ‘AI slop.’ Steven Bartlett’s media company, FlightStory, has already changed course, stopping AI-written LinkedIn posts after concluding authentic, human-written updates performed better.
TikTok is navigating the tension between its own AI ambitions and content quality. The platform requires creators to label AI-generated content, applies invisible watermarks to the metadata of some videos, and in November introduced a toggle letting users control how much AI content appears in their feeds. In July, TikTok said it was beginning to test improved detection systems to target accounts dedicated to posting AI-generated spam. Much of the company’s tech is built around ByteDance’s AI video model, Seedance 2.0.
Meta has taken the most permissive stance, allowing AI-generated content such as the viral Shrimp Jesus to circulate freely while adding ‘AI info’ labels across Facebook, Instagram, and Threads to content it detects as AI-generated. In June, Meta extended that labelling to advertisements using AI, whether created by Meta’s own tools or third-party services.
With Snap’s algorithm change taking effect now, the practical question for creators is whether the remaining platforms follow with harder enforcement or whether voluntary disclosure becomes the dominant model. TikTok’s planned detection upgrade may offer the clearest test of whether technical tools can keep pace with the volume of AI output they are trying to identify.