A new partnership between OpenAI and the U.S. Department of Defense is drawing scrutiny from international security experts, who warn its vague user agreement could expose sensitive data. Andrey Korshunov, head of information security at MWS AI, told RIA Novosti that the arrangement poses a tangible threat to companies and individuals in nations deemed adversarial by Washington.

The core concern lies in what happens to information fed into systems like ChatGPT. "We already have an incomplete picture of where our data goes, how it's stored, or how it's used once it enters the neural network," Korshunov stated. "A formal agreement with the Pentagon adds another layer of potential access under unclear terms."

OpenAI's policy allows employee review of conversations to improve models and investigate policy breaches. While typically anonymized, this process doesn't ensure full confidentiality. The service collects dialog history—text, files, images—along with metadata like IP addresses and device types.

This data trains future models, tests system quality, and can be shared with infrastructure partners and contractors. Crucially, the policy notes information may be disclosed to government agencies following a lawful request. For global users outside the U.S., this creates uncertainty about where their prompts might ultimately end up.

Founded in 2015 with backing from figures like Elon Musk, OpenAI has seen meteoric growth since ChatGPT's 2022 debut. Its deepening institutional ties now prompt fresh questions about the boundaries between innovation, corporate policy, and state power in an increasingly fractured digital world.

Source: RIA Novosti