Choosing a provider
Open storage at an AI aggregator: 144,000 chat logs
In October 2026 it emerged that 302.ai, a service offering ChatGPT, Claude, Gemini and other models in one place, had kept 17.6 million files, including more than 144,000 chat logs, in publicly accessible cloud storage.
302.ai is an AI aggregator run by a Hong Kong company: in a single interface, users can reach ChatGPT, Claude, Grok, DeepSeek and Gemini, among others. On 2 August 2026 Cybernews researchers found publicly accessible Microsoft Azure storage belonging to the service.
It held 17.6 million files totalling 22.72 TB. Among them were more than 144,000 chat logs with users' prompts, the models' answers, the chosen model and timestamps; some also included dates of birth, home addresses and contact details. Alongside them were images, videos, audio files and more than 26,000 archives holding conversations, summaries and files users had uploaded for analysis, including business documents.
The researchers notified the company on 11 August, and public access was closed by 8 September. According to the researchers, there is no evidence that anyone else downloaded the data, though earlier access cannot be ruled out. The case became public in October.
What can we learn from it?
An aggregator is convenient because it offers several models in one place, but our conversations are then also stored in another company's systems, and that company's security matters too. The large models are also available directly, through their makers' own apps and subscriptions. In Confiella's copy mode, the confirmed, pseudonymised request goes into your own ChatGPT, Claude or Gemini subscription, and the program explains the limits of this beforehand. Otherwise too, check who stores the conversations, and enter client data only in pseudonymised form.
Sources
- eSecurity Planet: 302.ai Data Leak: Private AI Chats Among 17.6 Million Exposed Files
- Cybernews: AI aggregator leak exposed sex doll support chats and 18 million files
This article is based on the public sources listed. If you find a mistake, write to us: [email protected]