"China, Did You Steal Our Technology?"... OpenAI and MS Launch DeepSeek Investigation

Chinese Startup DeepSeek Develops Low-Cost, High-Performance AI
OpenAI and Microsoft: "We Witnessed Attempts to Steal OpenAI Data"

Bloomberg News and others reported on the 29th (local time) that OpenAI and Microsoft (MS) have launched an investigation into whether Chinese AI startup DeepSeek unlawfully used OpenAI's data.

DeepSeek developed and showcased an AI model with performance comparable to OpenAI's ChatGPT at a low cost, raising suspicions that this was made possible through unauthorized collection of OpenAI's data.

OpenAI stated that it has witnessed multiple attempts by China-based entities to extract large amounts of data from its AI tools. It added that these attempts appear to be aimed at training DeepSeek's AI model using outputs from OpenAI's models.

MS security researchers also reported that last fall, individuals believed to be associated with DeepSeek tried to collect data via OpenAI's application programming interface (API).

OpenAI said it has banned the suspicious accounts and, together with MS, identified the parties behind the attempts.