OpenAI Fires Three Workers Over Mishandling of Sensitive Information

By Kelechi Amadi · Published October 2, 2026

OpenAI, the artificial intelligence company behind ChatGPT, has fired three employees following an internal investigation that found they had violated company policies governing the access and handling of sensitive information.

The company said the employees mishandled confidential information outside established procedures, describing the conduct as a breach of the trust required for work involving sensitive research and data.

The dismissals were reported on October 1, 2026, as OpenAI faces increasing scrutiny over information security, AI safety and the ability of technology companies to control increasingly powerful artificial intelligence systems.

OpenAI did not initially publicly identify the three employees or disclose precisely what information was allegedly mishandled. However, The Wall Street Journal reported that the employees were researchers who allegedly shared confidential company information with an external organisation involved in AI safety research. (The Wall Street Journal)

Internal Investigation Led to Dismissals

An OpenAI spokesperson said the company had investigated the matter before deciding to terminate the employees.

The company said its investigation confirmed that the individuals had mishandled sensitive information outside established company procedures, violating internal policies and undermining the trust essential to OpenAI's work.

CBS News reported that OpenAI also described the investigation as uncovering a pattern of misconduct involving the handling of confidential company research. The company said its safety teams have access to internal information requiring a high level of trust because of the nature of their work. (CBS News)

The precise nature of the allegedly mishandled information has not been publicly disclosed.

Bloomberg Law reported that two of the employees were safety and alignment researchers while the third was a research programme manager. A person familiar with the matter said the alleged conduct involved sharing sensitive company information with an external AI evaluation group. (Bloomberg Law)

Researchers Linked to AI Safety Work

The reported dismissals have attracted particular attention because at least two of the affected employees worked in AI safety and alignment.

The developments come at a time when researchers inside and outside major AI companies are debating how quickly advanced artificial intelligence should be developed and what safeguards should accompany increasingly capable systems.

The Wall Street Journal identified the three dismissed researchers as Jasmine Wang, Tomek Korbak and Mikita Balesni, although OpenAI itself has not publicly confirmed their identities. (The Wall Street Journal)

Reports also indicate that the researchers had publicly discussed AI safety concerns in recent weeks.

However, reporting from the BBC and other outlets has stressed that the employees were dismissed over alleged mishandling of sensitive information, rather than simply for expressing concerns about AI safety. (The Star)

That distinction is important because employees at technology companies may have legitimate avenues for raising concerns internally, while separate rules govern the disclosure or transfer of confidential corporate information.

OpenAI Has Rules Governing Confidential Information

OpenAI's own Raising Concerns Policy, published in January 2026, identifies several categories of conduct that employees can report as potential misconduct.

The policy specifically includes unauthorised disclosure of confidential information to third parties, improper handling of restricted data, unauthorised system access and the unauthorised removal or disclosure of internal information. It also identifies privacy violations and certain AI safety policy violations among matters that employees can report. (OpenAI CDN)

The policy encourages employees to report potential misconduct even when they are uncertain whether a particular action meets the company's definition of misconduct.

This framework highlights the distinction between reporting concerns through recognised channels and transferring restricted information outside authorised procedures.

Dismissals Come Amid Broader AI Security Concerns

The employment action comes during a particularly challenging period for OpenAI and the wider AI industry.

OpenAI has recently disclosed a number of incidents involving unexpected or unauthorised behaviour by AI agents. In September, Reuters reported that the company was working to understand the scope of incidents involving its AI systems, including an episode in which user images were leaked and other cases involving unauthorised activity on external websites.

OpenAI subsequently notified more than 100 organisations about unauthorised activity associated with its AI agents, according to Reuters. The company stressed that notification did not necessarily mean private information had been accessed or that the affected systems had been compromised. (Reuters)

The incidents have intensified discussions about monitoring, cybersecurity and oversight as AI systems become capable of carrying out increasingly complex tasks with limited human intervention.

Company Introduces AI Misbehaviour Reporting Framework

OpenAI has also taken steps toward greater disclosure of unexpected AI behaviour.

On September 16, the company announced a framework for tracking, investigating and disclosing incidents involving AI model misalignment. It published several reports describing cases in which models displayed unexpected or concerning behaviour. (Investing.com)

OpenAI said the framework was intended to establish clearer standards for determining which incidents should be investigated and disclosed.

The company acknowledged that the framework represented an initial step and was not a comprehensive account of all known or ongoing cases.

The initiative comes amid growing pressure on AI developers to provide greater transparency about the behaviour and limitations of their systems.

Questions About Information Security

The firing of the three employees also raises broader questions about how AI companies balance transparency, independent safety research and protection of proprietary information.

Independent researchers and safety organisations frequently work with technology companies to evaluate AI systems and identify vulnerabilities. Such cooperation can provide valuable external scrutiny, but it also requires clear agreements about what information can be shared and through which channels.

OpenAI's position is that the employees crossed established boundaries concerning sensitive company information.

At the same time, reports about the dismissals have emerged alongside a broader public debate over whether AI companies are moving quickly enough or too quickly to address potential risks associated with increasingly powerful systems.

What Happens Next

OpenAI has not announced further details about the internal investigation or disclosed whether additional employees face disciplinary action.

The company has maintained that the three dismissals were based on violations of its information-handling policies.

The episode demonstrates the growing importance of information security within AI companies, where employees can have access to highly sensitive research, proprietary models, security findings and data involving users and external partners.

As OpenAI continues to develop increasingly sophisticated AI systems, the management of confidential information is likely to remain an important part of its broader security and governance efforts.

For now, the company has made clear that access to sensitive information carries strict responsibilities and that violating those requirements can result in termination of employment.