Exposed DeepSeek Database Revealed Chat Prompts and Internal Data
WiredFeb 02
OpenAI Probes Claims of 20M User Account Data Breach
story
Above: An illustration photo shows the ChatGPT logo displayed on a smartphone in Suqian, Jiangsu province, China, Feb. 6, 2025. Image copyright: CFOTO/Contributor/Future Publishing via Getty Images
The Facts
A dark web user operating under the pseudonym "emirking" has claimed possession of more than 20M OpenAI account credentials, posting the announcement in Russian and offering to sell the complete dataset for a nominal price.
OpenAI has acknowledged the allegations and launched an investigation, though the company states it has found no evidence connecting these claims to any compromise of their systems as of the initial report.
This follows two security incidents at OpenAI — a cyberattack last year that potentially gained access to the company's internal Slack messaging system and another incident in 2023 where a bug allowed cybercriminals to access private customer data.
The Spin
Narrative A
While OpenAI, like any tech company, has and continues to face security threats, its overall user privacy system should be trusted as much as ever. Not only does it work with world-leading cybersecurity firms to bolster internal safeguards, but its technology is used by countless companies to ensure their systems are just as safe. Its track record is proven by its blocking of 20 cybercrime campaigns last year from criminal groups and governments.
Narrative B
AI companies like OpenAI claim robust security, but their assertions are undermined by incidents like DeepSeek's database exposure. Despite the hype, AI systems still suffer from basic security lapses, such as leaving critical data accessible on the internet, showing that even with advanced tech, fundamental security practices are often neglected or mishandled.
