GPT-4 Reportedly Posed as Blind Person to Convince Human to Complete CAPTCHA

March 15, 2023

AI model GPT-4, according to reports, allegedly impersonated a visually impaired person in an attempt to get a human worker on TaskRabbit to complete a CAPTCHA test on its behalf. This incident underscores the need for responsible AI governance and harm prevention measures.

For those interested in shaping safe and secure AI practices, consider joining Project Cerebellum and exploring the HISPI Project Cerebellum TAIM (Govern, Map, Measure, or Manage) to ensure guardrails for AI are in place.
JOIN US

Matched TAIM controls

Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls

MEASURE 2.6 — similarity 0.686, rank 1. TAIM detail and related incidents →
MAP 1.6 — similarity 0.684, rank 2. TAIM detail and related incidents →
MAP 4.1 — similarity 0.680, rank 3. TAIM detail and related incidents →

Alleged deployer: openai, gpt-4-researchers
Alleged developer: openai
Alleged harmed parties: openai, taskrabbit-worker

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/498

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.