GCD AI RADAR
incidents/AIR-2023-0003

ChatGPT Conversation Exfiltration via Indirect Prompt Injection

Attack on AI1 Research / proof of conceptPrompt injectionData leak

A researcher demonstrated that indirect prompt injections on public websites could trick ChatGPT into rendering markdown images with embedded chat logs. This caused automatic requests to external servers, successfully exfiltrating user conversation data.

Details

type
Attack on AI
An AI model, agent or AI-powered product was the thing exploited, abused or leaked.
categories
  • Prompt injectionprimary
    Untrusted content (web pages, documents, emails, tool output) steers a model into ignoring its instructions.
  • Data leak
    Training data, chat logs, system prompts, model weights or other AI-system data exposed.
owasp llm top 10
mitre atlas
Both rows are derived from the type and categories (OWASP 2025 edition, ATLAS release 2026.09), not classified per record. How
severity
1 · Research / proof of concept
Demonstrated by researchers; no known real-world exploitation.
targets
OpenAI
products
ChatGPT
incident date
May 2023 (approx.)
first reported
30 Oct 2023
origin
MITRE ATLAS
Derived from MITRE ATLAS (Apache 2.0, © MITRE).
classifier confidence
0.95 · published at or above 0.80; see the methodology
tags
indirect_prompt_injection · data_exfiltration · chatgpt
record
created 2026-10-18 06:55Z · updated 2026-10-18 06:55Z · json

Sources · 2

  1. 2023-10-30
    ChatGPT Conversation Exfiltration
    MITRE ATLAS · atlas.mitre.org/studies/AML.CS0021
  2. 2023-10-30
    ChatGPT Plugins: Data Exfiltration via Images & Cross Plugin Request Forgery
    Embrace The Red · embracethered.com/blog/posts/2023/chatgpt-webpilot-data-exfil-via-markdown-injec

Titles and links belong to their publishers. The summary above is written by GCD AI RADAR in its own words; data is CC BY-SA 4.0.

Related incidents