How Anna Makanju orchestrated OpenAI’s political charm offensive

Less publicized, though, was the driving force behind that world tour: Anna Makanju. But there she was in D.C., sitting over Altman’s left shoulder during congressional testimony in May; in New Delhi, speaking, alongside Altman, to India’s G20 Sherpa, Amitabh Kant; and in Seoul, taking part in a round table discussion alongside South Korean president Yoon Suk Yeol, OpenAI co-founder Greg Brockman, and yes, Altman. 

As OpenAI’s vice president of global affairs, Makanju has played an instrumental role in crafting OpenAI’s political charm offensive, which began long before Altman’s global excursion. In early 2022, Makanju orchestrated a demo for U.S. lawmakers that showed how they could use GPT-3 to quickly parse unwieldy pieces of legislation like the America COMPETES Act, a bill that promotes investments in innovation and stretches thousands of pages long. That same year, before OpenAI’s image generator Dall-E launched, she took the company’s chief scientist, Ilya Sutskever, on a trip to Washington to show off the technology for lawmakers. 

“I really had this vision of being a technical partner to regulators and ensuring that by the time that this technology was ubiquitous, they were ready,” Makanju says. 

Whether Washington is really ready for what OpenAI has unleashed is still up for debate. But at the very least, Makanju’s strategy appears to have put OpenAI in an entirely different political position than other tech giants, who lawmakers have treated less as collaborators, and more as punching bags. 

Born in Russia to a Ukrainian mother and Nigerian father, Makanju spent nearly a decade working in national security and foreign policy during the Obama administration. She served as the National Security Council’s director for Russia the year it invaded Crimea and later became then-Vice President Joe Biden’s special advisor for Europe and Eurasia. That work was some of the most rewarding of her career, Makanju says, but when the Trump administration took over and her position as a political appointee ended, she was looking for another role that would offer the potential to have a global impact. That search soon led her to Facebook, where she worked on global elections and policy. 

When Makanju joined Facebook in late 2018 — years after the 2016 election and months after the Cambridge Analytica scandal broke open — the relationship between the company and global policymakers had already curdled, leaving little room for anything but intermittent public floggings of Facebook executives on Capitol Hill. By the time she left the company to join OpenAI in September 2021, it had grown even worse, as Facebook was being blamed for fueling lies that the 2020 election was stolen.

Makanju describes her work with OpenAI as “night and day” from Facebook. In her current role, Makanju has tried to position the company as an early partner in the effort to mitigate the risks of generative AI, while also advocating for its potential benefits. She’s also helped spearhead a new industry-wide self-regulatory group called The Frontier Model Forum, which includes representatives from OpenAI, Google, Microsoft, and Anthropic, all of whom are coming together to work on safety standards for the next generation of large language models. “We really want to make sure that our best thinking on safety is being done jointly and is accessible to everyone,” Makanju says.

Makanju knows better than anyone that efforts by OpenAI to shape the regulations that govern it will inevitably be viewed skeptically. After all, companies like Meta and Google have spent years lobbying in their own self-interest, even while talking the talk about supporting regulation. Some have already argued that OpenAI’s calls for a licensing regime for AI systems will only entrench its power.

But Makanju insists OpenAI’s unique structure—it operates as a “capped-profit” company, which is governed by a nonprofit—is designed to keep the company honest and true to its mission of ensuring that the future of artificial intelligence ultimately benefits humanity, not just its executives. But, she acknowledges, “that may still not satisfy everyone.”

For Makanju — having both worked in government and for a tech company with an infamously combative relationship with government — there’s no question that a more cooperative approach to tech regulation is more beneficial for everyone. “Because we started this conversation sooner, we are still in a space where we can collaborate,” she says. “This is such complex technology, and it touches so many sectors, it’s really critical that the conversation remains collaborative.”

This story is part of AI 20, our month-long series of profiles spotlighting the most influential people building, designing, regulating, and litigating AI today.