OpenAI Terminates Three Researchers Over Mishandling of Sensitive Information Amid Intensifying AI Safety Debate

Introduction

The artificial intelligence industry has entered a period of unprecedented internal scrutiny, with leading laboratories facing mounting pressure to reconcile rapid technological advancement with robust safety governance. Employees at several high-profile AI firms have increasingly spoken out publicly about existential risks associated with the development of ever-more-powerful systems — a trend that has drawn both public attention and internal consequences. Amid this backdrop of OpenAI researcher terminations and growing discourse around AI safety policy, the San Francisco-based ChatGPT developer has confirmed the dismissal of three of its researchers for alleged violations of company information-handling protocols. Reportedly, OpenAI’s response to this incident reflects the company’s stated commitment to maintaining internal trust and procedural integrity at a time when the broader industry is navigating significant reputational and regulatory crosscurrents.

OpenAI Officially Confirms the Termination of Three Research Employees Following Internal Policy Investigation

OpenAI has formally confirmed the separation of three employees following an internal investigation into the mishandling of sensitive company information. The San Francisco-based AI laboratory, developer of the widely used ChatGPT platform, stated that its investigation concluded the individuals had shared or transferred sensitive information outside of established company procedures, constituting a breach of internal policy.

“We have parted ways with three individuals,” OpenAI stated in a formal communication issued to AFP. “Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work.”

OpenAI did not officially confirm the identities of the three terminated employees in its public statement. However, reporting by the Wall Street Journal and Bloomberg identified the researchers as Jasmine Wang, Tomek Korbak, and Mikita Balesni. According to both outlets, at least two of the three individuals worked on safety and alignment — among the most sensitive and closely watched research disciplines within the organisation. The terminations are understood to involve activities connected to an external organisation that conducts evaluations of artificial intelligence models, though OpenAI has not disclosed further operational details regarding the nature of the alleged breach.

The Terminated Researchers Worked Primarily on AI Safety and Alignment, Roles Central to OpenAI’s Core Research Mission

Reportedly, the three researchers whose employment was ended — Jasmine Wang, Tomek Korbak, and Mikita Balesni — had each been publicly vocal on matters related to AI safety in the weeks preceding their dismissal, particularly on the social media platform X.

According to reporting by the Wall Street Journal, all three individuals had regularly posted content addressing AI safety-related concerns. Mikita Balesni posted on September 10 that he believed artificial intelligence carried more than a 10 percent probability of causing the deaths of all humans, a statement that echoed sentiments expressed by other employees across multiple leading AI laboratories in recent weeks.

Tomek Korbak addressed the growing public discourse directly in a post dated September 11, writing: “I’m quite unhappy with much of what OpenAI does. I am very happy that I’m allowed to say ‘I’m quite unhappy with much of what OpenAI does.'” Jasmine Wang, for her part, responded to a public resignation statement by former Anthropic researcher Jacob Coxon, warning that “it’s hard to overstate how dangerous speeding towards RSI is” — a reference to recursive self-improvement, a technique in which software systems are designed to continuously teach and enhance themselves without direct human intervention.

The substance of their public statements places all three squarely within a broader movement of AI researchers who have raised alarms about the pace and direction of frontier model development.

OpenAI’s Researcher Dismissals Emerge Within a Broader Industry Context of AI Safety Concerns and Voluntary Regulatory Commitments

A segment of the AI research community has long grappled with the tension between organisational loyalty and the perceived moral obligation to disclose safety-related concerns to the public. The OpenAI researcher terminations arrive at a particularly charged moment in this ongoing debate, as employees at multiple leading AI laboratories have become notably more outspoken about risks they associate with the rapid development of advanced artificial intelligence systems.

The week before these dismissals became public, a 27-year-old researcher named Jacob Coxon announced his resignation from Anthropic — a competing AI laboratory — with a direct warning that leading AI labs, including OpenAI where he had previously been employed, were “gaming with our lives” by pursuing an accelerated path toward increasingly powerful models. Coxon’s resignation statement catalysed a wave of public commentary from researchers across the industry, including the posts by Wang and Balesni cited in subsequent reporting.

According to publicly available information, leading US technology companies reached a voluntary agreement this week to implement self-regulatory measures on AI safety, following a meeting with President Donald Trump at the White House. President Trump characterised the agreement as a “morally binding” commitment to construct adequate safeguards around the technology. Executives representing Nvidia, Google, Meta, xAI, OpenAI, and Anthropic all signed the agreement — a development that underscores the degree to which AI safety governance has become a central concern for both industry leadership and the executive branch of the US government.

OpenAI Has Operated as a Leading AI Research Organisation Since 2015, With Its Internal Trust Policies Now Subject to Heightened Public and Regulatory Scrutiny

Public records show that OpenAI was founded in San Francisco in 2015 and has since grown into one of the most prominent artificial intelligence research laboratories in the world, developing products including the widely adopted ChatGPT conversational AI platform. The organisation employs hundreds of researchers and engineers across disciplines including safety, alignment, policy, and model development.

The company’s stated internal policies require that sensitive information be handled exclusively within established company procedures — a standard the terminated researchers were found to have violated, according to OpenAI’s own investigation. The incident has drawn attention to the structural tensions inherent in AI safety research, where employees are simultaneously expected to identify and internalise risks and to manage how and when such concerns are communicated externally.

Industry observers note that the firings follow a period during which OpenAI has faced multiple separate controversies, including a US state-level investigation into a reported rogue AI incident in August 2026 and litigation brought by Elon Musk alleging that the company had abandoned its founding mission. These developments collectively reflect the heightened regulatory and reputational environment in which major AI laboratories now operate.

OpenAI’s headquarters are located in San Francisco, California. The company’s official communications channel for press inquiries is press@openai.com, and its primary public-facing platform is openai.com.

Frequently Asked Questions About OpenAI’s Termination of Three Researchers Over Sensitive Information Mishandling

Why did OpenAI fire three researchers? OpenAI terminated three researchers — identified by the Wall Street Journal as Jasmine Wang, Tomek Korbak, and Mikita Balesni — after an internal investigation determined that all three had mishandled sensitive information outside of established company procedures, violating internal policies and, in OpenAI’s own words, “breaking the trust essential to our work.”

Who are the three researchers OpenAI fired? The three individuals terminated by OpenAI are Jasmine Wang, Tomek Korbak, and Mikita Balesni, according to reporting by the Wall Street Journal and Bloomberg. OpenAI did not officially confirm their identities in its public statement, but at least two of the three worked on safety and alignment research within the organisation.

What sensitive information did the OpenAI researchers mishandle? OpenAI has not disclosed the specific nature of the sensitive information involved. The company’s statement indicates that the individuals transferred or shared sensitive materials outside established internal procedures, and that the conduct was connected to activities involving an external organisation that evaluates artificial intelligence models.

Were the fired OpenAI researchers involved in AI safety work? According to the Wall Street Journal and Bloomberg, at least two of the three terminated researchers worked in AI safety and alignment — research disciplines focused on ensuring that advanced AI systems behave reliably and do not pose undue risks to humans.

What had the fired researchers said publicly before their termination? All three researchers had posted publicly on X in the weeks before their termination. Mikita Balesni stated on September 10 that he believed AI carried more than a 10 percent probability of killing all humans. Tomek Korbak expressed dissatisfaction with OpenAI’s direction on September 11. Jasmine Wang warned against the dangers of recursive self-improvement, a technique where AI software is designed to enhance itself continuously.

What is recursive self-improvement, which Jasmine Wang warned about? Recursive self-improvement (RSI) is a technique in AI development where software systems are designed to continuously teach and improve themselves without direct human intervention. Critics, including Wang, have argued that accelerating toward RSI without adequate safeguards poses serious and potentially catastrophic risks to human safety.

Did OpenAI participate in the US government’s voluntary AI safety agreement? OpenAI was among the companies whose executives signed a voluntary AI safety agreement following a meeting with President Donald Trump at the White House. President Trump described the commitment as “morally binding,” and signatories included executives from Nvidia, Google, Meta, xAI, OpenAI, and Anthropic.

Closing Summary

The dismissal of three researchers by OpenAI — confirmed by the company in a formal statement and further detailed by the Wall Street Journal and Bloomberg — marks a significant moment in the ongoing tension between AI laboratory governance and the growing movement among AI safety researchers to speak publicly about perceived existential risks. OpenAI’s internal investigation found that Jasmine Wang, Tomek Korbak, and Mikita Balesni mishandled sensitive information in violation of established company procedures, actions the company characterised as a fundamental breach of organisational trust. The incident unfolds against the backdrop of a newly signed voluntary AI safety agreement between the US government and major technology companies, reinforcing that the balance between transparency, safety advocacy, and institutional accountability remains one of the defining challenges of the current AI development era.

For more information on OpenAI’s official statements and research activities, readers may contact:

• Website: openai.com
• Press inquiries: press@openai.com
• Headquarters: OpenAI, San Francisco, California, United States