Worth Knowing
Freepik Becomes Magnific — and Adds Prohibitions
Freepik — the stock-asset-and-AI-tools company — is rebranding to Magnific, the name of one of its own acquired AI products. The change is uniform across the company's tracked documents, with a continuity banner across the top of each: "Freepik is now Magnific. Your account, projects, files, subscription, and 250M+ stock assets come with you."
The substantive change is in the AI Products section of the Terms of Use, where two new prohibitions have been added...
to the restricted-use list:
- Deploy subliminal or deceptive techniques, for instance deepfakes, with the purpose or the effect of materially distorting a person's behavior in a way that causes or may cause harm.
- Generate, manipulate, distribute, or otherwise use content whose creation, possession, or dissemination is prohibited under applicable criminal law, including content that involves serious harm, abuse, or exploitation of individuals or groups.
Both clauses track EU AI Act Article 5 language almost verbatim. The deepfake prohibition is new and explicit; the criminal-content clause is broader than the existing categories in Magnific's Acceptable Use Policy and is now incorporated directly into the Terms of Use.
Why this matters for AI content producers: Two things. First, the practical one — the deepfake prohibition is now an enforceable contract term on a major image-generation platform, carrying the same legal weight as the rest of the Terms of Use. If you're producing AI-generated likenesses through Magnific tools, this is the language that governs the work. Second, the contrast is worth noting: in the same scrape window, Meta deleted six similar Article 5-aligned prohibitions from its global Meta AI Terms (see The Big One above; whether they survive in regional documents we don't yet track is an open question). Working from the same regulatory text, Magnific and Meta made opposite drafting choices on the global versions of their documents.
In human terms: A freelance designer is asked by a sketchy client to "tweak" a real public figure's face into an embarrassing scenario for what the client calls a harmless prank. The freelancer was already declining on ethical grounds, but the client kept pushing because no platform rule expressly prohibited it. Now the freelancer can point to Magnific's new clause 10 — explicit deepfake/subliminal-manipulation ban — as a contract-level shield.