OpenAI employees warned about operational security lapses for years, sources say

For years, OpenAI employees have been warning about operational security lapses, sources reveal. Even before the release of ChatGPT, some staff members raised concerns about lax security practices internally, according to five people familiar with the situation. They claimed that management was more focused on keeping up with competitors like Google and Anthropic rather than prioritizing security.

Employees argued that outsiders, including foreign governments, could easily obtain sensitive information about OpenAI's technology due to its lax controls. One former employee accused of planting a recording device is said to have Saudi backing, as per court filings.

In response to these internal warnings and employee complaints, OpenAI has begun to shift its strategy to address security issues. Thibault Sottiaux, a senior executive overseeing product development, recently took charge of an internal initiative to secure OpenAI's internal systems. Additionally, a quarter of OpenAI's production engineers were reassigned to improve the company's security infrastructure in August.

However, the tension within the AI field is evident, as AI technology has become a strategic asset for the US, putting pressure on American companies to protect their technology from foreign rivals. This has led to competing interests between security and the rapid advancement of AI technology, potentially complicating OpenAI's and Anthropic's plans for massive initial public offerings.

Will DePue, a former OpenAI researcher, expressed this view, stating that with competition so intense, AI leaders "can't sacrifice velocity for security."