---
title: "Researcher's 2 million planted canaries expose AI data privacy leaks"
url: https://www.parallelquant.com/posts/researcher-s-2-million-planted-canaries-expose-ai-data-privacy-leaks-631f6e
source_name: "Ars Technica"
source_url: https://arstechnica.com/security/2026/11/a-leaky-ai-supply-chain-is-exposing-secret-user-data-at-an-alarming-rate/
published: 2026-11-20T20:59:08.000Z
topics: ["security"]
publisher: "Parallel Quant"
---

# Researcher's 2 million planted canaries expose AI data privacy leaks

*2026-11-20 · Source: [Ars Technica](https://arstechnica.com/security/2026/11/a-leaky-ai-supply-chain-is-exposing-secret-user-data-at-an-alarming-rate/)*

A researcher seeded about 2 million prompts containing unique, trackable secrets into the AI supply chain to see where the data would surface. The test showed that data submitted to AI systems does not stay private.

**Why it matters:** Canary testing turns vague privacy worries into measurable evidence about where user data flows across model providers, routers and third-party tools. It lands as more products route requests across multiple models and retain chats longer, as with Google keeping Gemini chats even with history off, increasing the number of places data can leak. This kind of audit could become a template for regulators and enterprise buyers.

**Topics:** security

---
Read the original: https://arstechnica.com/security/2026/11/a-leaky-ai-supply-chain-is-exposing-secret-user-data-at-an-alarming-rate/
Canonical: https://www.parallelquant.com/posts/researcher-s-2-million-planted-canaries-expose-ai-data-privacy-leaks-631f6e
Published by Parallel Quant — https://www.parallelquant.com
