Open AI strengthens security in response to a series of wiretapping and hacking... “Employee warning ignored”

Open AI, the developer of ChatGPT, has belatedly begun to reorganize its security system following a series of security incidents, including the discovery of a wiretapping device in a conference room and a series of hacking incidents.

Reuters reported on the 3rd (local time), citing multiple sources, that OpenAI has begun a project to strengthen its internal security system led by TiVo Sotio, head of product and platform.

Sotio General Manager reallocated a quarter of product engineers last August to improve security infrastructure.

Chief Information Security Officer (CISO) Dane Stuckey is also introducing regular security checks on internal tools and taking steps to limit access to sensitive research content.

The reason OpenAI is strengthening security like this is because it was recently revealed that Saudi Arabian engineer Abdulrahman Alfozan had installed a wiretapping device in the company's conference room, and security incidents such as messenger hacking continued.

The source said that Open AI employees have repeatedly warned the company and the board of directors that the company has neglected internal security due to its focus on speeding up artificial intelligence (AI) development.

For example, most of the internal code was stored in a single repository called a 'monorepo', so any employee with access to the repository could access the details of most of the projects taking place within the company.

There were also many cases where key research documents were widely shared on external platforms such as GitHub or Slack.

Tools such as the 'sandbox' (an AI model verification environment that blocks the external environment) that verifies new AI models were also not subject to regular security checks.

Because of this, some employees claimed that “security controls were so lax that external hackers or foreign governments could easily steal the company’s core AI secrets.”

Former researcher Will Defew, who left OpenAI in April, pointed out to social networking service (SNS)

In response to these comments, OpenAI spokeswoman Elana Wiedman explained that OpenAI is investing heavily in security programs, saying, "OpenAI is one of the most heavily targeted organizations in the world."