Steven Bartlett, host of the podcast The Diary of a CEO, asked whether AI companies were "gaslighting the public in a way." Karen Hao said they were. She then asked him a question: "if most of the climate scientists in the world were bankrolled by fossil fuel companies, do you think we would get an accurate picture of the climate crisis?" Bartlett said no. Hao said AI is already in that position, because "the AI industry employs and bankrolls most of the AI researchers in the world."
Hao is a journalist who covers AI's effects on society. She wrote Empire of AI, a book investigating the rise of OpenAI and the race to build ever more powerful AI. Their exchange is in a "most replayed moment" clip of The Diary of a CEO, published October 23, 2026, and taken from an earlier full-length conversation. Hao makes her case in three steps. She starts with who pays for AI research, then gives one example from Google and one from OpenAI.
Two kinds of influence
Hao said the companies monopolize "knowledge production." In her account, they present themselves as the only people who really understand how the technology works. That lets them treat public dislike of AI as ignorance: if people object, she said, the message is that they "don't actually know enough about this technology." She said the companies take this line with both the public and policymakers, and that they have "captured the majority of the scientists" who study what AI can and cannot do.
Hao described two ways that money shapes the field. The first is soft agenda-setting. Companies "set the agenda on AI research in soft ways, simply by funneling money to their priorities so that only certain types of AI research are produced." Nobody has to forbid a question for it to go unstudied, because work that nobody funds tends not to get done. The second is more direct: the companies "will censor researchers when they do not like what the researcher has found."
The Google case
Hao's example of censorship involves Timnit Gebru, who co-led Google's ethical AI team. Gebru, Hao said, "was literally hired to critique the types of AI systems that Google was building." Gebru co-wrote a critical paper on large language models, the technology behind chatbots such as ChatGPT. Hao said the paper showed how these models were leading to certain kinds of harm. She said that "in an attempt to try and stop this research from being published," Google fired Gebru and then her fellow co-lead, Margaret Mitchell.
Reuters reported that Gebru said her December 2020 firing followed her challenge to an order not to publish the paper. Google gave a different reason when it dismissed Mitchell, who had co-authored the paper, on February 19, 2021. The company said she had moved confidential material outside the company, and it cited an investigation it had begun on January 19. Reuters also reported that Google announced changes to the way it reviews research before publication.
The knock on the door
Bartlett then recalled a video of Hao's in which a young man said someone from one of the big AI companies came to his door and asked for his emails and text messages. Hao said the company was OpenAI, which had started subpoenaing some of its critics. A subpoena is a legal order to hand over documents or to testify. She said it looked like "a campaign of intimidation," and also like "phishing for more information" to "map out the network of critics further."
The man, Hao said, runs a small watchdog nonprofit. Watchdog groups "like Midas," meaning the Midas Project, had been questioning OpenAI's plan to convert from a nonprofit into a for-profit company. Hao said that at the time, finishing the conversion was "sort of existential" for OpenAI. In her account, the watchdogs wanted transparency and public debate about a move she called unprecedented, and were trying to keep it from happening "in the dead of night." She said that was when he was served papers.
According to Hao, the papers demanded every communication he had had that might involve Elon Musk. She blamed "the strange paranoia that OpenAI had that Musk was somehow funding these people to block the conversion." She said none of them were funded by Musk, and that he simply replied that he had no such documents because they did not exist.
The man Hao described is most likely Tyler Johnston, who founded the Midas Project. On September 3, 2025, the San Francisco Standard reported that OpenAI had issued two subpoenas dated August 12 as part of its litigation with Musk. One went to the Midas Project and one to Johnston personally. The requests went beyond what Hao described. They covered communications involving Musk or Mark Zuckerberg, the identities of funders, donation details, and documents about OpenAI's governance and restructuring. Johnston told the paper that Midas had received no money from Musk or Meta and that he had not discussed the project with Musk. He said that because he was the group's only full-time employee, producing the documents could eat up his working time. OpenAI's lawyer Anne O'Leary defended the requests as an effort to establish transparency about funding.
Hao said OpenAI ultimately succeeded in converting. OpenAI's own October 28, 2025 announcement calls it a recapitalization. Under it, the nonprofit, renamed the OpenAI Foundation, keeps control of the commercial business, which is now a public benefit corporation called OpenAI Group PBC. OpenAI valued the Foundation's stake in that business at about $130 billion.