Pinterest puts AI slop on blast with new labels

If you have used Pinterest recently looking for real home decor ideas or makeup inspiration, you have probably noticed your feed getting cluttered with glossy, unrealistic AI images, many of them leading to shady third-party sites.

Following several months of initial testing, Pinterest has officially rolled out labels for AI-generated content to all users worldwide, aiming to curb the wave of AI junk hitting the platform.

Automatic detection over self-reporting

When you tap into a close-up view of a Pin, an "AI modified" tag will now appear in the bottom left corner if the image contains generated or altered elements.

Crucially, Pinterest is not relying on creators to voluntarily disclose whether they used AI tools. Instead, the platform is combining automated classifiers with photo metadata inspection to flag content, even when there are no obvious visual tells.

On the technical side, Pinterest is leveraging the IPTC Photo Metadata Standard. While usually used for alt-text descriptions, IPTC data also tracks editing processes and creation tools. For cases where the automated system misses the mark, Pinterest has set up an appeals process for creators who believe their work was incorrectly labeled.

A "see fewer" button for beauty and art

Tagging images is only half the battle. Pinterest is also testing an option that lets users actively filter AI imagery out of their feeds.

Users will soon see a "see fewer" option inside the three-dot overflow menu in the bottom right corner of a Pin. Selecting it signals the algorithm to downrank and show less AI-generated content in that feed.

Pinterest is rolling this test out first in categories hit hardest by synthetic media, specifically beauty and art, with plans to expand it to other verticals later.

Why it matters

For everyday pinners, this is a desperately needed tool to protect utility. People use Pinterest to plan real-world projects, buy clothes, and try recipes. Flooding feeds with impossible AI-generated aesthetics breaks that contract, especially when spammers use synthetic pins as clickbait for scams.

Across the social landscape, platforms have mostly relied on honesty badges that bad actors simply ignore. Pinterest taking a proactive stance by combining metadata standards, computer vision classifiers, and negative recommendation signals puts pressure on rival visual platforms like Instagram to step up their moderation.

We think giving users a direct "see fewer" lever is the right move, though execution will make or break it. The line between heavy photo editing and generative AI can be blurry, and if legitimate photographers get hit with AI tags repeatedly, Pinterest will have a creator revolt and a massive appeals backlog on its hands.