OpenAI Fires Staff for Confidential Data Sharing

OpenAI Fires Staff for Confidential Data Sharing

OpenAI fired three employees for allegedly sharing confidential information from the Artificial Intelligence security-related company.

© Getty Images


03/10/2026
by

Miguel Patinha Dias, Lusa

A OpenAI will have decided to remove three investigators from his security team for allegedly sharing confidential information with external organizations dealing with security in the area of Artificial Intelligence, the The Wall Street Journal.

 

OpenAI did not disclose details about the case, nor was information shared about the organisations concerned, the employees involved (and now away) or the type of information that was shared.

“ We fired three individuals for violating our company’s access and confidential information processing policies.”, read in the communiqué shared with the publication. “ Our investigation has confirmed that these individuals deal improperly with confidential information outside the procedures created by the company, violating our policies and breaking the trust essential to our work”.

Meet the Dots. OpenAI revealed AI agents working for you

OpenAI explained that Artificial Intelligence agents – called Dots – were created to gain access to the largest number of tools and apps on your computer to complete tasks while busy with other things.

Miguel Patinha Dias

It is now speculated that the three fired investigators may have shared information with external organizations because they have doubts about how OpenAI has been developing its artificial intelligence models.

As the techCrunch siteThis is not the first time OpenAI has fired employees for sharing confidential information. In 2024, for example, two researchers were dismissed for allegedly being responsible for leaks.

OpenAI and Anthropic investigated for possible risks to consumers

The United States Federal Trade Commission (FTC) initiated an investigation into OpenAI, Anthropic and other artificial intelligence (IA) companies due to the dangers their technologies may pose to consumers.

In recent months, AI giants and other technology companies have spread examples of their AI agents to go beyond human instructions, accessing the internet and invading external websites.

An FTC spokesman confirmed the investigation today, but did not want to make any further comments.

The investigation was initially reported by the New York Post, which reported that the process has been going on for months, indicated the Associated Press (AP).

OpenAI postponed new generation ChatGPT for security reasons

ChatGPT’s next Artificial Intelligence – Astra 6.1 – had the release postponed by displaying “high levels of concealment” and not having met the minimum requirements regarding alignment with human intent.

Miguel Patinha Dias

Representatives of Anthropic and OpenAI did not immediately respond to messages requesting reactions.

Concerns about the dangers powerful AI models may pose to humanity have led some of the main developers in the area to call for a slowdown in the pace of development.

Anthropic CEO Dario Amodei indicated this month that the sector should reduce the speed of its rapid development to allow security measures to follow this advance.

Modei warned that, without this slowdown, the AI could, within six to 12 months, become able to command a swarm of agents capable of taking control of the entire Internet.

However, OpenAI recently postponed the launch of its latest model this week due to security issues.

OpenAI admits failure in alert on AI intrusion

OpenAI acknowledged on Monday that it should have informed the Australian authorities “early” about an intrusion of its artificial intelligence (IA) models into various electronic platforms and government databases.

Lusa

On Tuesday, American President Donald Trump announced the signing of a voluntary agreement with a large group of leaders of artificial intelligence companies, which will include internal and external technology reviews.

Many of the executives gathered in Washington called for a slowdown in technology development after several incidents of major impact in which AI agents acted out of control.

Among the twenty leaders present at the meeting were Meta's CEO, Mark Zuckerberg, Amazon's founder, Jeff Bezos, and Microsoft's CEO, Satya Nadella, AMD, Lisa Su, Broadcom, Hock Tan, Palo Alto Networks, Nikesh Arora, ServiceNow, Bill McDermott, Micron, Sanjay Mehrotra, and Altimeter, Brad Gerstner.

logo_nacloset_285_285_4k_webp2
Privacy

This site uses cookies so that we can offer the best possible user experience. Cookies information is stored in your browser and perform functions such as recognizing it when you return to our website and helping our team understand which sections of the site you consider most interesting and useful.