VSCO Bars Creator Content From Generative AI Model Training

By Amara Osei

VSCO says it will not use public or private Creator Content to develop, train or improve generative AI models, whether users pay for membership or use the service without paying. The revised Creator Content Standards make that commitment explicit and apply the same restriction to outside AI providers.

Why did VSCO revise its wording?

VSCO came under scrutiny in May after its Terms of Use appeared to authorize broad uses of user content. One clause granted VSCO rights to the name, image, voice or likeness of any individual in a user's content.

The rights extended to content in whole or part, in any form or media, and to technology known or developed later. The terms also gave VSCO the right to use certain content for Creator Promotion and to develop, train and improve AI or machine learning models under the Creator Content Standards.

That wording suggested a path for using photographers' work in generative model development, which prompted alarm. VSCO quickly said it would not train AI models on users' content, but the response did not settle concerns raised by the clause.

The company then sent a letter to users explaining the change. Eric Wittman framed the issue as a question of trust, saying users deserved a straight answer about what happened to their work.

The letter warned that sharing work should not create uncertainty about use in systems generating photos or video. The revision replaces interpretive risk with an explicit prohibition.

What does the updated standard cover?

The updated standards state that public or private Creator Content will not be used to develop, train or improve AI or machine learning models for generative AI use. This applies to paid members with active subscriptions and to non-paying creators.

Wittman stated the policy directly: “We don’t use your content to train generative AI models.” He added that the rule covers public and private work and applies equally to paying and non-paying users.

For photographers, the relevant change is operational rather than a new feature. Under the revised standard, privacy setting and payment status do not determine whether submitted work may train a generative system, removing two variables from platform selection and client workflow decisions.

Wittman tied the clarification to VSCO’s history: “For the past 15 years we have championed photographers.” He said the company built a business around helping photographers make their work and earn from it, not treating photographs as raw material for models that generate images, video or other media.

This revenue framing matters because paid members and free creators receive the same training restriction. VSCO presents the policy as a baseline platform term, not a premium benefit tied to subscription revenue.

Does VSCO still use AI?

VSCO does. The platform continues to use machine learning technology and non-generative AI across its services. Wittman said those systems help photographers find work, search for people to connect with, edit images and get more from VSCO products.

Wittman also said the photography community had identified those applications as useful. The source does not identify individual systems or assign a named product to each function.

Here, generative AI means systems that generate new photos, video or other media. Non-generative AI refers to operational machine-learning tools that support discovery and editing without using user photographs to train content generators.

That distinction keeps the announcement relevant to daily workflow. Photographers may still rely on a platform that uses non-generative AI for discovery and editing, while receiving an explicit prohibition on training generators with submitted work.

The announcement does not name the non-generative systems or explain the technical controls enforcing the separation. It therefore describes the company’s intended boundary, not a technical inventory of automated features.

How does the restriction extend to outside providers?

VSCO says external AI providers powering its features have already agreed not to use content submitted through the service to train generative AI models. The company says it holds those providers to the same distinction it applies to its own systems.

Wittman acknowledged that the earlier Creator Content Standards were too broad to make the distinctions and intentions clear. He said users should not have to decipher a policy to learn what happens to their work.

The provider commitment extends the stated protection beyond VSCO’s own infrastructure, where content may pass through third-party services. Still, the announcement names no providers, publishes no agreement terms and describes no audit, monitoring system or penalty for breach.

On the evidence presented, photographers have a stated safeguard rather than independently verified proof. For business use, the update narrows a material concern about generative training without quantifying enforcement risk.

What should photographers watch next?

The revised Creator Content Standards are available on VSCO’s website. Wittman also framed the policy in photographers’ authorship, describing work as evidence of time spent learning, places visited, people trusted and choices only the photographer could make. He said the photographer’s perspective and decisions remain their own.

The announcement declares no new fee, royalty or revenue-sharing terms. It describes a policy clarification rather than a commercial launch.

Photographers should watch whether VSCO preserves this precise separation as external AI services and platform features expand. The practical benchmark is narrow: operational tools may remain, but submitted work must not become training material for systems that generate new media.