Pinterest is giving users a volume knob for AI — sort of. This week, the company quietly rolled out a new setting that lets people reduce how much generative-AI imagery shows up in specific parts of their feed, but it stops short of promising a clean, all-human stream. The new control, described by Pinterest as an updated “tuner,” targets the kinds of categories that have been swamped by what users and journalists have called “AI slop” — think beauty, art, fashion and home décor — and it’s available now on Android and desktop, with iPhone promised in the coming weeks.

The tuner lives in Settings → Refine your recommendations under a “GenAI interests” tab. From there, you can tell Pinterest you’d like to see less GenAI content in the categories you care about. But there are two important caveats: the tool applies only to “eligible image Pins,” and it’s limited to categories Pinterest has identified as especially prone to AI modification. In short, it’s targeted and incremental — helpful if your “home décor” feed has gone surreal, less useful if you want a blanket ban on anything AI-flavored.

Pinterest frames the tuner as giving users more control over personalization: pick the categories, tweak how much AI you want to see, and the company says it’ll expand category options over time based on feedback. The move is part product control, part damage control — an attempt to stop AI-generated images from weakening Pinterest’s core promise of practical, actionable inspiration.

This isn’t the first time Pinterest has tried to fend off an AI flood. In late spring, the site started adding “AI modified” labels to images its systems flagged as generated or edited using generative AI. The label appears when Pinterest detects AI use via image metadata (following IPTC standards) or when its in-house classifiers say an image looks like GenAI content. That earlier move was meant to increase transparency; the tuner is the next step — letting people reduce their exposure rather than just notice it.

Industry writers have been blunt: users called the problem “AI slop” and warned that if ordinary inspiration boards turn into feeds of impossible clothes and unmanufacturable furniture, Pinterest’s usefulness — and brand trust — could erode. Some outlets even framed it as a broader existential moment for social platforms that sell themselves as human-first inspiration engines. Pinterest’s own communications lean hard on reassurance, saying the changes are designed to strike “the right balance between human creativity and AI innovation.”

The hard part: detecting AI — and getting it right

Detection is the fulcrum here, and detection is messy. Pinterest’s approach uses metadata when present (some AI tools embed provenance fields) and machine classifiers tuned to spot AI artifacts when metadata isn’t available. That approach is necessary — many bad actors use local or modified generators that leave no metadata — but it’s also imperfect. False positives and strange edge cases have already surfaced: creators reporting that perfectly legitimate, hand-made photos and original artwork have been labeled “AI modified,” sometimes without a quick or clear route to appeal. For sellers, photographers and small creators, a mistaken AI tag can mean lost sales or credibility.

Why Pinterest isn’t promising a total opt-out

Two pieces explain the practical reason for that: first, not all Pins are “eligible” for the tuner; second, some AI content is indistinguishable from human-made images once it’s been lightly edited or stripped of metadata. TechCrunch noted that Pinterest’s product team wants to avoid giving users a false sense of control if the underlying classifiers can’t absolutely guarantee identification in every case. In other words, the company can reduce your exposure in targeted areas, but it can’t guarantee a 100% AI-free Pinterest until provenance and detection become reliably perfect at scale.

What creators and users should watch for

If you’re a creator: check your Pins for the GenAI label and, if necessary, use Pinterest’s appeals channels and metadata best practices (embed clear provenance where possible). If you’re a user fed up with impossible dresses and uncanny interiors, the tuner is a practical tool to experiment with; toggle it, check how your feed changes, and send feedback when Pins still feel off. Pinterest says it will add categories over time and take user signals into account, so the tuner’s practical effectiveness will depend on iteration and data.

Pinterest’s tuner is a recognition that platforms built around human curation can’t ignore the tidal rise of generative imagery. It’s a pragmatic, product-forward move that acknowledges user frustration without pretending the problem can be instantly solved. For now, the best way to read Pinterest’s signal is this: the company is listening — enough to build a knob — but the work that will actually restore trust (more accurate detection, faster appeals, provenance standards) is still ongoing. Until those pieces are in place, “dial down” is a welcome but limited fix.