OpenAI researchers sent a letter to the board of directors warning of a powerful AI discovery that they said could threaten humanity, according to two people familiar with the matter. The letter and the algorithm behind it, a project internally called Q* (pronounced Q-Star), were cited as one factor that led to CEO Sam Altman’s ouster last week, the sources said. Reuters, which broke the story, reports that the letter predated Altman’s firing but was not the sole cause.

The Q* project reportedly shows a new ability to solve simple math problems, a task that today’s large language models generally fail at because they lack true reasoning. Given vast computing resources, the model aced grade-school-level math, making researchers optimistic about its future. Some inside OpenAI view the breakthrough as a potential step toward artificial general intelligence — systems smarter than humans. OpenAI declined to comment on the report.

The revelation has ignited intense speculation in AI circles about what Q* actually is and whether it represents a genuine leap forward. Altman himself, speaking at the APEC summit the day before his firing, said that “just in the last couple of weeks” the company had pushed “the veil of ignorance back.”