Week in AI: Lina Khan comes to YC, OpenAI shuffles safety lead

FTC Chairwoman Lina Khan spoke on a panel in San Francisco hosted by Y Combinator.

Tom Williams

By Sara Bloomberg - Staff Reporter

July 25, 2024, 02:56pm PDT

Washington and local officials paid a visit to Y Combinator in San Francisco on Thursday for a conversation about artificial intelligence, regulations, startups and the technology landscape in California.

YC hosted an event featuring FTC Chair Lina Khan and Jonathan Kanter, the assistant attorney general in the Dept. of Justice's antitrust division. They were joined by YC CEO Garry Tan, California State Senator Scott Weiner, and representatives from the Mozilla Foundation, Electronic Frontier Foundation and Omidyar Network.

Openness was a major theme: Open markets, open source software, open models and open talent pools.

"It can be tricky knowing how to strike the right balance, and again, policymakers do have a lot legitimate concerns," Khan said. "That said, I think it would be a real missed opportunity if we didn't position openness to really get a fair shot in this marketplace, especially from a competition perspective."

Senator Weiner also expressed his own openness to feedback from the tech and startup community regarding his proposed AI regulations which are being considered in his controversial Senate Bill 1047 that is advancing through the state legislature. He tried to allay fears that the bill would be too stringent or drive tech companies out of California, particularly in light of Meta's decision to not release the multimodal version of its Llama large language model in Europe due to regulatory concerns.

"This bill is so much more narrow than European Union's AI bill," Sen. Weiner said, and noted that the proposed bill isn't about where a company is located but, rather, whether its AI models are deployed in California.

Aleksander Madry speaks at O'Reilly AI Conference 2019 on April 17, 2019 in New York City. (Photo by Ben Hider)

Ben Hider

Departures

OpenAI has reportedly reassigned an executive who led the organization's "preparedness framework" on assessing and mitigating "catastrophic risks." Aleksander Mądry is taking on a "bigger role" within OpenAI, the Information reported, though it was unclear what prompted the move and what the new role would be.

"This is wrong. Aleksander is working on a new and (very) important research project," CEO Sam Altman wrote on X in response to the Information's report. "Aleksander will continue to support preparedness work in various ways."

Nonetheless, the change comes after two other high-ranking safety leaders made exits earlier this year. OpenAI co-founder and former Chief Scientist Ilya Sutskever resigned in May, followed just hours later by his fellow "superalignment" and safey co-lead Jan Leike.

Reports at the time said that OpenAI was decentralizing the company's safety-related efforts but was still committing 20% of its compute towards similar work.

Legal Watch

Condé Nast – which publishes Wired, the New Yorker and Vogue magazines — sent Perplexity AI a cease-and-desist letter, telling the San Francisco startup to stop using the publisher's content in its AI search results, the Information reported. In June, Perplexity also waded into hot water with Forbes which accused the startup of stealing its paywalled content.

Funding rounds to know

Canadian AI startup Cohere raised a $500 million Series D round that more than doubled the company's valuation to $5.5 billion and brings its total funding to $970 million, Bloomberg News reported on Monday. By Tuesday, CNBC reported that Cohere had laid off around 20 employees from its workforce of around 400 people. The company, which is developing enterprise AI products, is based in Toronto and also has offices in San Francisco and London.