Luanti removed from Google Play due to baseless AI copyright notice

TL;DR Highlight

A vague AI-generated DMCA notice got the Luanti app pulled from Google Play

Who Should Read

A backend developer building AI-driven detection or moderation pipelines who worries about false positives

Core Mechanics

Luanti's app was pulled by a DMCA notice
- Tracer.AI filed on behalf of Microsoft, claiming infringement of Minecraft.
- An update at the top says Microsoft withdrew the notice and the app is back on the Play Store.
The notice had no specifics
- It cited only the US registration of Minecraft Java Edition 1.9.
- It did not say which assets were allegedly used.
Luanti ships almost no assets
- It is a game-creation platform with no games by default.
- Minetest Game was removed from the bundle in December 2023.
It has happened before
- A similar notice from the same company in March 2023 took 46 days to resolve after a counter-notice.
- In February, the indie game Allumeria was temporarily removed from Steam and Microsoft later dropped the notice.
Tracer.AI uses AI agents for detection

Community Reactions

Jurisdiction inconsistency noted
- A commenter noted the notice claims Vanuatu jurisdiction while other recent claims used the US.
- They asked whether this might be fraud.
No cost for false positives
- One commenter said false positives cost the sender nothing while false negatives lose revenue.
- Another proposed requiring a bond that pays damages if a notice is reversed.
Speculation about AI-similar code
- One commenter hypothesized AI trained on open source might produce similar code that a scanner then flagged.
- This is not confirmed.
Alternative install routes and a counterpoint
- Commenters pointed to F-Droid and Obtainium and urged diversifying away from Google Play.
- One outsider argued the screenshots look like Minecraft, so 'baseless' is overstated.

How to Apply

Add a human approval step to automated enforcement
- If your pipeline auto-blocks or reports, require human review before the final action.
- Today, list every step that runs without a human.
Make evidence fields mandatory in detections
- Require structured fields for the target asset and similarity evidence, and reject empty ones.
- Do not pass vague results that only cite a registration number.
Prepare a distribution path outside the store
- If you ship an app, set up a non-store distribution channel in advance.
- Document the takedown response process and contacts.

Try It

copy & run

# Paste into Claude or ChatGPT — a simplified checklist prompt based on this case

Our service uses AI to automatically detect {{target}} and then performs {{action: delete/block/report}}.
Current flow: {{describe your pipeline}}

Review it on these points.
1. Harm and recovery time when a detection is wrong
2. Where a human must confirm before action
3. Concrete evidence fields that must accompany every action
4. How to respond to a counter-notice

Put the result in a table and name the single riskiest point first.

Next: add the suggested evidence fields to your real schema.

Terminology

DMCAA US copyright takedown process. Platforms remove reported content to avoid liability.

counter-noticeA formal objection to a wrongful notice. The law says content should be restored 10 to 14 business days later.

perceptual hashingA fingerprinting technique for image similarity. It finds look-alikes even when files are not identical.

Tracer.AIA company that uses AI agents to detect brand infringement and send takedown requests.