One sentence decided which videos get shown
Snap published a short post on its newsroom on 31 July under the title Rewarding Authentic Creativity on Spotlight, and one sentence in it does all the work: wholly AI-generated videos will no longer be eligible for recommendation on Spotlight. The company dated the change to the current month and going forward. Spotlight is Snapchat's public feed of vertical video, the surface where a clip reaches people who do not already follow the account that made it, which makes recommendation eligibility the difference between an audience and an archive.
The post carried one number worth holding on to. Snap said the count of unique Spotlight contributors globally is up more than 120 percent compared with last year. That is the pressure behind the rule. A feed that doubled its supply of contributors in twelve months has to choose what to surface, and Snap has now written one of those choices down rather than leaving it inside the ranking model.
The exemption is where the real rule lives
Read the next sentence and the policy changes shape. Content that has been enhanced or edited using Snapchat's AI creative tools will continue to be eligible for recommendation, and Snap adds that such content will include transparency indicators. So the presence of artificial intelligence in the production chain is not what disqualifies a video. A clip can be AI-touched and stay eligible, provided the AI in question is Snapchat's.
That is the part no headline carried, and it is the part an operator should write down. The variable being tested is not synthesis, it is provenance. Two videos that look identical on the screen can land on opposite sides of the line depending on which company's software produced them. Snap has not drawn a boundary between human work and machine work. It has drawn one between tools inside its own product and tools outside it, and then labelled the inside of that boundary authentic.
Eligibility was never a promise of payment
A good deal of the coverage described this as Snapchat ending payment for AI videos, with headlines about banned monetisation and stopped rewards. Snap's post does not mention monetisation, rewards, or payment anywhere. It speaks only about eligibility for recommendation. The distinction is not pedantry, because eligibility has never been a guarantee of anything: a video that is eligible for recommendation may still be recommended to nobody, earn nothing, and disappear.
Getting this right changes what the news actually is. Nothing contractual was withdrawn, because nothing contractual was there. What happened is smaller and, for anyone planning a content operation, more consequential in the long run than a payment rule would have been. A platform has introduced a question about your toolchain into the machinery that decides your reach, and it did so in a document that reads like a statement of values rather than a change of terms.
Disclosure satisfies the law, not the feed
European teams now answer to two tests that were designed to do different jobs. Under the EU AI Act, supervision and enforcement moved into force on 2 August 2026, with the AI Office and national authorities responsible for implementing and enforcing the regime; the obligations in that direction are built around telling the audience what it is looking at. Labelling is the currency. Do it properly and you have met the duty the law imposes on you.
Snap's rule is indifferent to labelling. It does not ask whether the creator disclosed the use of AI, and disclosing it does not restore eligibility for a wholly AI-generated clip. A marketing team in Berlin or Madrid can therefore be fully compliant with European transparency duties and still be structurally invisible on the feed, having satisfied a regulator and failed a classifier on the same file. These are not two versions of one rule. They are two rules with different owners, different tests, and no reciprocity between them.
Treat your creative stack as a distribution dependency
The practical response is an inventory, and it is narrower than a policy review. For every channel that matters to you, record which software produced the asset, not merely whether AI was involved, because that is now the question being asked. Where a platform privileges its own generative tools, using them is a distribution decision rather than a creative preference, and it should be priced as one alongside the lock-in it quietly creates. Snap also said plainly that no detection system is perfect, which is a warning about false positives on work you made yourself, with no visible appeal.
The wider pattern is the one to plan against. When a platform exempts its in-house tools from a restriction it applies to everyone else's, it is standing on both sides of the market it regulates: it sets the rule, sells the compliant option, and judges the results. Expect that shape to spread, because it is cheap to announce, hard to audit from outside, and it moves creators onto the platform's own software without a single commercial negotiation.
Read next: The Only Binding AI Check Left Is European | Two AI Rules Landed Today, One Lets You Check



