Hacker Accessed Details on OpenAI Technology Last Year

A hacker successfully infiltrated OpenAI’s systems in an apparent effort to steal details about its technology, the New York Times reported in early July.

The hacker could not access the code of the various technologies that OpenAI is building but did access the company’s internal messaging system. From there, the hacker acquired details of OpenAI technology by accessing messages shared between various developers and teams.

OpenAI believes the incident is the work of a private individual and not tied to any foreign government. As such, OpenAI did not deem it a threat to national security and did not contact the FBI or other law enforcement agencies.

The details of the hack, which took place early last year, were shared internally with OpenAI staff members at an all-hands meeting in April 2023 and with the Board of Directors. Nothing was communicated to the general public, as no customer or partner information was exposed.

Questions have been raised recently about whether OpenAI is focusing enough on safety and security, though the company claims it remains steadfast in its commitment to safety.

OpenAI recently formed a Safety and Security Committee to evaluate current safeguards and propose recommendations for improvements if needed. This committee was put in place a few months after OpenAI dissolved its Superalignment team, a safety-focused research group.

OpenAI was also one of 16 companies that came together at a global summit in May and pledged to develop its AI technology safely. The pledge has the various companies committing “to publishing safety frameworks for measuring risks, to avoid models where risks could not be sufficiently mitigated, and to ensure governance and transparency.”