Adobe stops utilising user content for AI training

Adobe has made it clear in its amended terms of service that it will no longer use user content-both locally saved and cloud-stored-to train its generative artificial intelligence (AI) systems. Only material uploaded to Adobe Stock for Adobe Firefly training is exempt from this rule. Chief Strategy Officer Scott Belsky stressed that the goal of this explanation is to clear up any misunderstandings from the past.

The revised agreement now comprises several distinct categories, one of which is dedicated to generative AI. Adobe's Chief Strategy Officer, Scott Belsky, clarified that these changes do not represent a shift in policy but are intended to clear up confusion.

Belsky said, "We've explicitly said we will not train generative AI on your content. It was always a policy that we had as a company. We always made that very clear, but we never explicitly said that."

The updated terms also address concerns surrounding Adobe scanning content, created under a nondisclosure agreement (NDA). The company said it does not "scan or review" work stored locally on user devices. Automatic scanning is only applicable to content uploaded to the cloud, to "ensure we are not hosting illegal or abusive content, like Child Sexual Abuse Material."

Human review of cloud-uploaded work occurs if it's flagged as illegal or if users opt into a prerelease, beta, or product improvement program. The recent changes in the terms of service were misinterpreted by users as granting Adobe permission to use their work for AI training, leading to widespread frustration among creatives.

The Department of Justice (DOJ) is currently suing Adobe for allegedly hiding costly cancellation fees, and making it difficult for users to cancel their subscriptions.

Belsky acknowledged that some user frustration could be traced back to Adobe's shift to a subscription-only model in 2012.