Skip to content
Amogh Suman

An AI platform that reviews apps for dark patterns

A review platform that checks apps and websites against all 13 dark pattern categories in India's 2023 guidelines, used to review more than 25 enterprise clients.

Sector
Financial services and consumer apps
Scope
Apps and websites of 25+ enterprise clients
My role
Product lead: architecture and delivery

The problem

India's 2023 dark pattern guidelines name 13 practices that apps must avoid. Checking a large app against all 13, screen by screen, takes a team days and still misses things.

The Guidelines for Prevention and Regulation of Dark Patterns, issued in 2023 under the Consumer Protection Act, list practices such as false urgency, basket sneaking, drip pricing and subscription traps. Every app that sells to Indian consumers has to stay clear of them.

Reviews used to be manual: someone walks through each journey, takes screenshots and judges each screen against the list. Results depended on who did the review, and a second pass rarely matched the first.

The approach

  1. Map every journey

    Break each app into its journeys (discovery, purchase, payment, renewal, cancellation, consent withdrawal) so that no screen is judged out of context.

  2. Capture screens and copy

    Collect screenshots and on-screen text for each step, including fine print, pre-ticked boxes and countdown timers.

  3. Classify with AI against all 13 categories

    A model checks each screen against definitions written from the guidelines and returns the category, the evidence on screen and a severity.

  4. A person confirms every finding

    Reviewers accept or reject each flag, so the final report rests on human judgement while the AI does the first pass.

  5. Report as a register and a matrix

    Findings land in a risk register and a matrix of journeys against categories, colour coded by status, with the screenshot behind every cell.

Who it serves

  • Compliance and legal teams

    A defensible view of where the app stands on each of the 13 categories.

  • Product and design teams

    The exact screens to change, with the reason for each.

  • Leadership

    One matrix that shows risk across products and journeys, and how it moves over time.

The outcome

The platform has reviewed the apps and websites of more than 25 enterprise clients, faster and with more consistent findings than a manual pass.

It also changed the conversation with clients. Instead of a list of opinions, teams receive evidence for every finding and a matrix they can track release after release.