OpenAI accused of alleged global AGI bidding war plan
In a recent interview, Leopold Aschenbrenner, former safety researcher at OpenAI, made startling allegations about the company's plans for artificial general intelligence (AGI). He claimed that OpenAI had once planned to initiate a global AGI bidding war among the governments of China, United States, and Russia. "At some point several years ago, OpenAI leadership had laid out a plan to fund and sell AGI by starting a bidding war," Aschenbrenner stated during his conversation with tech podcaster Dwarkesh Patel.
Aschenbrenner's dismissal from OpenAI: The inside story
Aschenbrenner also revealed the reasons behind his termination from OpenAI earlier this year. He was reportedly dismissed after the company's human resources department took issue with a memo he had sent to senior management. In this memo, he warned about the potential theft of "key algorithmic secrets" by the Chinese Communist Party. According to Aschenbrenner, his warning was labeled as "racist" and "unconstructive," leading to questions about his loyalty to the company.
OpenAI's response to Aschenbrenner's allegations
In response to Aschenbrenner's allegations, OpenAI has stated to Business Insider that it shares his "commitment to building safe AGI," but disagrees with his portrayal of the company's work. The company has not yet commented on the specific allegations regarding a planned global AGI "bidding war." Aschenbrenner had previously shared documents with outside researchers during a brainstorming session, which HR deemed confidential and considered as a leak. This incident also reportedly contributed to his dismissal from the company.
Aschenbrenner's concerns over OpenAI's alleged AGI plans
Aschenbrenner expressed his astonishment that OpenAI would consider "selling AGI to the Chinese and Russian governments." He drew attention to the potential implications of initiating a bidding war and then playing these governments off against each other. Patel responded to these allegations by stating, "That's pretty f***** up." The company has yet to respond directly to these specific claims.
