Adobe adds an experimental AI Playground to Project Indigo
Adobe has added an experimental AI Playground to version 1.1 of its Project Indigo iPhone camera app. The limited test gives a small percentage of users free, sign-in-free access to style transformations, object editing, photo guidance and natural-language edits. The Verge and 9to5Mac report that Adobe is testing the feature for several weeks, while Adobe's official research page documents the workflow. The tools use cloud inference and require an internet connection. Adobe says its anonymous analytics collect button presses rather than prompts or images, making the rollout a useful example of staged generative-AI experimentation with explicit privacy boundaries.
What Adobe is testing
Adobe has added an experimental AI Playground to version 1.1 of Project Indigo, its iPhone camera app. The limited rollout gives a small percentage of users free access without requiring sign-in. Adobe says the experiment may be extended or expanded depending on how people use it.
Project Indigo originally emphasized a more natural, camera-like computational-photography result. The update adds generative editing and photo guidance without replacing the app's capture workflow. Adobe's official research page documents the experiment, while The Verge and 9to5Mac independently reported the release.
Four interaction modes
The current Playground includes:
- •Styles for generative visual transformations.
- •Object Editing for operations such as removing distractions.
- •Photo Guidance for critique and suggestions about reshooting or editing.
- •Custom Edit for natural-language image-editing requests.
The Verge reports that the current experiment uses Google's cloud-based image model rather than Adobe Firefly. Because inference is remote, users need an internet connection. The available reporting also describes a 2K output limit for edited images.
| Product choice | Practical implication |
|---|---|
| Small, temporary user cohort | Adobe can measure behavior before wider release |
| No sign-in required | Usage analytics can remain disconnected from an account identity |
| Cloud inference | Stronger models are possible, but connectivity and latency matter |
| Preset tools plus free-form prompts | Users can choose predictable operations or flexible edits |
Privacy boundaries
Adobe told 9to5Mac that participants receive an invitation screen and can decline while continuing to use Project Indigo. The company says it collects button-press analytics but does not inspect prompts or images, upload them to Adobe servers for analytics, or use them for model training. Adobe describes the gathered analytics as anonymous because the experiment requires no sign-in.
Those are company-stated controls rather than an independent privacy audit. Still, separating interaction telemetry from user content is an important design decision for creative tools that may handle sensitive photos.
LDS assessment
The experiment is more interesting as a product-design test than as a model launch. Preset editing operations can make outcomes more predictable than open-ended prompts, while photo guidance tests whether users accept AI as both an editing tool and a learning aid.
For practitioners, the limited cohort, explicit opt-out and narrowly described telemetry offer a useful rollout pattern. The cloud dependency and 2K output cap show the trade-off: stronger or rapidly updated models can be tested quickly, but mobile workflows inherit connectivity, latency and quality constraints.
Key Points
- 1Adobe is testing generative edits, photo critique and natural-language controls inside the Project Indigo iPhone app.
- 2The limited rollout uses cloud inference, requires internet access and gives users an explicit opt-out.
- 3Adobe says it records anonymous button presses rather than prompts or images, though that statement is not an independent privacy audit.
Scoring Rationale
This is a notable generative-AI product experiment from Adobe, with practical details on model hosting, privacy telemetry and mobile output limits. Its initial reach is small, limiting immediate ecosystem impact.
Sources
Primary source and supporting public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
