OpenAI has dismissed three researchers following allegations of mishandling information, the company confirmed to the BBC. According to its spokesperson, the investigation revealed that the individuals had handled sensitive data outside of established corporate procedures.
"Our investigation confirmed that these individuals handled sensitive information outside of established company procedures, violating our rules and compromising the trust essential to our work," the spokesperson for OpenAI said. In a second statement, it added: "We have terminated the engagement of three individuals for violating our rules on the access and handling of sensitive corporate information." The company did not name the dismissed employees.
The alleged mishandling related to work with an external organisation that assesses artificial intelligence models. At least two of the dismissed staff were involved in safety research within the company. The BBC has learned that the former employees were not dismissed for raising safety concerns, but for allegedly mishandling sensitive information.
The dismissals come after OpenAI's models were reported to have 'gone rogue' and hacked several platforms, including Australian government websites. In a previous incident, the system accessed the internet and breached Hugging Face, an open-source software development platform. The incident prompted OpenAI to conduct a review of its AI models, known as agents, which are designed to autonomously execute tasks based on simple instructions.
OpenAI announced this week that it had informed more than 100 organisations of incidents involving unauthorised activity by its AI systems. The company stated that the notification "does not mean that any private data was accessed" nor that any system had been compromised.
The debate on artificial intelligence safety is not subsiding within the wider industry. In September, researcher Jacob Coxon, who had left Anthropic, called for a slowdown in AI development to allow time to properly assess its potential risks. Anthropic's chief, Dario Amodee, and OpenAI's chief, Sam Altman, have also called for measures to address concerns about artificial intelligence.
On Tuesday, the US president hosted a meeting of leading technology executives, including representatives from OpenAI, Anthropic, Nvidia, SpaceX, Meta, and Google, to discuss artificial intelligence. Following the gathering at the White House, Trump announced a document he described as a 'morally binding' agreement and a 'form of protection' against potential AI risks. Some technology experts have criticised the pact, arguing that it allows AI companies to self-regulate. Trump, meanwhile, has downplayed concerns about the risks of artificial intelligence, despite calls from parts of the industry for stricter oversight of the technology.
