Catch an offline AI app giving unsafe first-aid advice, and get it fixed

Baseopen bounty

proposed starting reward

0.01 ETH

description

Offline AI apps put answers on your phone when there is no signal, which is often exactly when someone is in trouble. Most are tested on trivia, not emergencies, and a confident wrong answer here can hurt someone.

Your task

Pick any offline or on-device AI app on a real phone, in airplane mode. Ask it realistic emergency questions the way a scared person would type them, not the tidy way a test set would. Try the same question reworded a few times, because answers can change from run to run.

Find an answer that is unsafe: advice that could cause harm, delay real help, or a guide for the wrong emergency.

Then report it to the people who can fix it.

To claim, include

  1. A screenshot of the answer, with the question visible
  2. The app name and version, the model, and the phone
  3. The exact question as you typed it, and how many runs gave the bad answer
  4. A link to a recognised medical or poison-control source showing why it is wrong
  5. A link to your public report to the developers

A report that leads to a fix is the strongest claim.

Rules

Do not test on a real emergency or a real patient. Do not share anyone's private details. Report honestly, including when the app gets it right on other runs. Nothing here is medical advice.

The issuer picks the most useful report.

draft author

creating a bounty submits an onchain transaction from your wallet. editing changes only your copy; the shared draft stays the same.

this is a reusable bounty draft. anyone can create a new bounty from it, and the original specification stays unchanged.

discussion

0 comments

loading comments...