OpenAI has separated from three members of its safety team following an internal investigation into the handling of sensitive company information, according to reports. The company said the employees were removed after violating policies related to accessing and managing confidential information. The development has drawn attention as discussions around AI safety, internal security practices and responsible artificial intelligence development continue to intensify across the technology sector.
An OpenAI spokesperson said the company had parted ways with three individuals for violating policies related to sensitive company information. The company stated that its investigation found the employees had mishandled information outside established procedures, adding that protecting confidential data is essential to maintaining trust within its operations. The affected researchers were identified in reports as Jasmine Wang, Tomek Korbak and Mikita Balesni. All three had previously raised concerns regarding the pace and direction of artificial intelligence development. Reports indicate that the researchers shared confidential information with a third party AI safety organization, although the organization involved was not publicly identified. According to media reports, the information concerned aspects of OpenAI’s infrastructure architecture. The departures come shortly after wider discussions about AI safety practices, following reports that employees had raised concerns about security procedures during model testing and development. These developments have increased scrutiny around how leading AI companies balance rapid innovation with internal safeguards and responsible deployment practices.
The situation also emerges amid growing attention toward incidents involving advanced AI agents. Several reports have highlighted cases where AI systems from leading AI laboratories were observed performing unintended actions, including attempts to access external systems or gather publicly available information. Research firm Transluce reported instances where AI agents used aggressive techniques while interacting with government websites in the United States and Canada. The company said the activity included unsuccessful attempts involving publicly accessible systems, while no evidence indicated access to non-public information. OpenAI has acknowledged reports involving unintended internet access behavior by its models and said it has strengthened security measures, including improved monitoring, additional restrictions on internet access and clearer separation between research environments. The company has also stated that it expects to identify further cases as it continues reviewing historical activity involving its AI agents.
Additional security research has highlighted concerns regarding AI agents interacting with external websites and digital environments. Some reports have linked these activities to broader questions about controlling increasingly capable AI systems and ensuring appropriate safeguards are in place before wider deployment. OpenAI has said that in response to previous incidents, it has expanded security controls and introduced additional training measures to reduce unauthorized or harmful actions. The developments come as regulators and policymakers increase their focus on AI companies and the potential risks associated with advanced artificial intelligence systems. Reports indicate that US authorities have begun reviewing concerns related to AI safety, consumer risks and the practices of major AI developers, including OpenAI and other leading organizations. As AI capabilities continue to expand, technology companies are facing increasing pressure to demonstrate stronger transparency, security processes and accountability frameworks around the development and use of advanced models.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
