Meta AI Bug in Deployed Service Reportedly Allowed Potential Access to Other Users' Prompts and Responses

December 26, 2024

A security researcher disclosed a potential access vulnerability in Meta AI's deployed chatbot service, which under specific circumstances could permit unauthorized users to view other users' prompts and AI-generated responses. The issue reportedly stemmed from guessable prompt IDs and insufficient server-side authorization checks. Meta swiftly rectified the problem in January 2025 and confirmed no signs of malicious exploitation, rewarding the researcher for their findings. It underscores the significance of robust governance and trustworthy AI practices. For those interested in shaping responsible AI governance, JOIN US to contribute to HISPI Project Cerebellum TAIM (Govern) efforts.

Learn more about how this incident maps to our Harm Prevention initiatives within the AI incident database.

Matched TAIM controls

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

MANAGE 4.3 — similarity 0.712, rank 1. TAIM detail and related incidents →
MEASURE 2.10 — similarity 0.706, rank 2. TAIM detail and related incidents →
MEASURE 2.6 — similarity 0.702, rank 3. TAIM detail and related incidents →

Alleged deployer: meta
Alleged developer: meta
Alleged harmed parties: meta-users, general-public

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/1172

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.