GCD AI RADAR
incidents/AIR-2020-0024

GPT-2 Model Replication by External Researchers

Attack on AI1 Research / proof of conceptModel theft / extraction

Researchers successfully replicated OpenAI's GPT-2 language model prior to its full release using publicly available technical information. The event highlighted risks related to tiered release schedules.

Details

type
Attack on AI
An AI model, agent or AI-powered product was the thing exploited, abused or leaked.
categories
  • Model theft / extraction
    Weights stolen, or model behaviour replicated through extraction or distillation attacks.
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, Open-weight model
products
GPT-2
incident date
22 Aug 2019
first reported
23 Oct 2020
origin
MITRE ATLAS
Derived from MITRE ATLAS (Apache 2.0, © MITRE).
classifier confidence
0.90 · published at or above 0.80; see the methodology
tags
model-extraction · replication · gpt-2
record
created 2026-10-18 06:55Z · updated 2026-10-18 06:55Z · json

Sources · 3

  1. 2020-10-23
    GPT-2 Model Replication
    MITRE ATLAS · atlas.mitre.org/studies/AML.CS0007
  2. 2020-10-23
    Wired Article, "OpenAI Said Its Code Was Risky. Two Grads Re-Created It Anyway"
    Wired · wired.com/story/dangerous-ai-open-source
  3. 2020-10-23
    Medium BlogPost, "OpenGPT-2: We Replicated GPT-2 Because You Can Too"
    blog.usejournal.com · blog.usejournal.com/opengpt-2-we-replicated-gpt-2-because-you-can-too-45e34e6d36

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