We score movies here. This week we're scoring a measuring stick, because the measuring stick is the more important release.
The Synthetic Motion Picture Ratings System launched at smprs.org and Screamify signed on immediately, making it the first streaming platform to rate its own catalog for AI content. Every Screamify Original gets a number. Our verdict: 8.4. The design is better than it needed to be, and the enforcement is worse than it should be.
Start with what the scale gets right, which is that it is a scale. Six tiers with real percentage bands, not adjectives. SMP-0 Practical-Only is 0 percent AI — live actors, real sets, human script, traditional post — and the system calls it the gold standard. SMP-1 caps at 15 percent. SMP-2 runs 15 to 35. SMP-3 runs 35 to 60. SMP-4 runs 60 to 85. SMP-5 Fully Synthetic runs 85 to 100 percent, digital humans and AI voices and machine-written scripts included.
Numbered bands are the entire innovation. Every AI-labeling scheme floated since 2024 has failed at the same place: it asked for a binary. Was AI used, yes or no. That question is now useless, because the honest answer for nearly every production above a microbudget is yes, somewhere, in the roto or the cleanup or the color. A binary label that catches everything catches nothing. Bands separate a film with an AI grammar pass on the script from a film with no humans in it, and those are not the same product.
The bands are also placed where the arguments actually are. SMP-3 is the interesting line: 35 to 60 percent, script co-written up to half by AI. That is the tier where authorship stops being a technicality, and the system puts a boundary exactly there rather than blurring it. SMP-4 keeping human principals while conceding majority-AI environments describes a real production model that already exists and had no name.
Now the number that matters for our score. Certification is self-submitted: register the title, attest the information is accurate, pay $29.99, receive the rating and the badge set. Premium tiers advertise enhanced AI-usage verification, but the floor is an honor system with a price tag. A ratings body that cannot audit is a ratings body that will eventually certify something wrong and have no mechanism to fix it. That costs a full point and a half here, and it is the only thing standing between 8.4 and a 9.
Adoption is the offsetting variable, and Screamify moved the needle on day one. A scale nobody uses scores a 4 no matter how well it's built. One platform committing its whole Originals slate makes it a live standard with real inventory behind it.
The credibility test came faster than expected, and the system passed it. Multiverse, Screamify's fully AI-generated interdimensional-cable channel arriving in October, is rated SMP-5 — the top of the scale, the purple badge, the maximum disclosure. The platform that launched the standard put its most synthetic property in the most synthetic tier. Every incentive pointed toward calling it a 3 and hoping nobody counted. They didn't.
The verdict: 8.4. A well-built scale, honestly applied by its first adopter, with an audit gap that needs closing before the badges start moving money. Watch whether a second platform signs on. That's the number that decides whether this is a standard or a website.
Comments (4)