Article title: Nearly 2 Million Requests Expose AI Privacy Flaw: "Zero Data Retention" Can Also Leak
Body:
Nearly 2 Million Requests Expose AI Privacy Flaw: "Zero Data Retention" Can Also Leak
IT之家, November 21 — Tech media outlet Ars Technica published a blog post today (November 21) reporting that HD Moore, Chief Research Officer at security firm Dragos, sent nearly 2 million requests containing trackable "bait" to multiple AI service providers, finding that the AI service chain has problems including hackers stealing prompts, resellers leaking data, and failures to comply with privacy settings.
The researchers deliberately sent a batch of trackable "test secrets" (not collected from real user channels) to AI, then observed whether any other programs or people later used them to access systems, thereby determining whether the data had leaked and which part of the service chain it might have leaked from.

The investigation found signs of compromise at at least three service providers, with attackers using stolen bait keys to launch tens of thousands of cloud service API calls, attempting to find more credentials, activate services, and maintain access. Some keys were used within hours of submission, and in some cases requests were rejected before the model even ran, yet the keys within them were still stolen.
The risk also comes from the multi-layered service chain. Many AI applications distribute requests through intermediaries, so content may pass through multiple platforms and be decrypted and processed at each stage. The research also found that some resellers log and transmit requests externally; for some requests marked "zero data retention" or "no data collection," their private URLs were still being scraped several days later.
Moore recommends that sensitive businesses consider local deployment; users who cannot self-host should choose trusted providers directly where possible and explicitly specify the upstream provider.

- Three breaches, endless leaks: Your AI supply chain needs canaries
Advertising statement: External links contained in this article (including but not limited to hyperlinks, QR codes, passcodes, and other forms) are used to convey more information and save selection time; the results are for reference only. All articles on IT之家 that contain external links include this statement.
Related Articles
- Gartner forecast: The world will see a net increase of about 1.7 million jobs due to AI by 2030
- Axelera AI unveils data-center-class AIPU chip Titania architecture, production to begin in 2027Q3
- Developer's self-made AI tool tracks indoor abnormal noises, pinpoints sound source to within about 40 centimeters
- iOS 27.3 will complete the Apple Siri AI ecosystem puzzle: understanding your health information, covering 92 types of data
- A U.S. law professor tried to use Anthropic to update lecture page numbers, but instead had a full year of class notes deleted by the AI
- Shizuku AI unveils full-duplex model Shizuku-Duplex, laying the foundation for AI companion development