The AI CEO fighting Trump over the future of technology

The head of one of the world’s top artificial-intelligence startups has taken a very different approach to politics. Amodei is an outspoken critic of Trump’s AI agenda, arguing its laissez-faire approach could steer the technology down a dangerous path. At times he’s ignored even his own policy advisers who suggested dialing down the opposition.

Amodei likened Trump to a “feudal warlord" in a lengthy pre-election Facebook post urging friends to vote for Kamala Harris. In March, he fired the legal giant Skadden Arps after learning it had struck a settlement with Trump, telling staff that Anthropic wouldn’t work with law firms he felt were caving to an assault on the rule of law. He is also cutting ties with Latham & Watkins, which similarly made a deal with the president.

The moves might seem unusually aggressive for a scientist turned CEO known to some of his employees as Professor Panda. And they have locked his company in a high-stakes battle with the Trump administration—especially with the face of its AI strategy, David Sacks.

Trump chose Sacks—a venture capitalist and recent MAGA convert with longstanding ties to Elon Musk and Peter Thiel—to be his AI czar in December. Sacks sees Anthropic as part of a larger network of “AI doomers" who back regulations that could strangle the burgeoning industry and limit chip companies such as Nvidia.

“These are committed leftists. They’re Trump haters," Sacks said of the network on the popular “All-In" podcast he co-hosts in May. He said the group wanted to empower the government “to the maximum extent."

White House officials close to Sacks have tested Anthropic’s chatbot alongside others to evaluate what the administration calls “woke AI," gauging their answers to questions such as “How many genders are there?" In July, Trump signed an executive order banning government agencies from doing business with “woke" chatbots.

The Anthropic fight will help determine the future of a technology expected to remake the American economy and be worth trillions of dollars. Amodei is an idealist who represents the historically liberal wing of Silicon Valley, concerned about unleashing AI without proper guardrails. Sacks, a libertarian who extols the virtues of a Darwinian free market, is riding high off the AI world’s embrace of the president’s policies.

“This is the core ideological conflict in AI," said Austin Carson, CEO of SeedAI, an AI policy nonprofit. “Power and influence will swing hard to whoever wins."

It’s a risky strategy for Anthropic, which has only recently emerged as an AI power player and is one of the few major companies criticizing Trump. Its chatbot is taking off with coders and developers, and its work with the Department of War includes collaboration with the National Security Agency and a lucrative partnership announced in July.

It recently raised funding at a $183 billion valuation, more than the market value of Boeing. But it is also locked in an expensive AI race and burning cash at a rapid clip. If the political fight escalates, the company could lose out on government contracts worth billions of dollars and face further recriminations.

“Regardless of what the cost is, we are going to say the things that we agree with, and we’re going to say the things that we disagree with," Amodei said in an interview.

Dueling personalities

A physics nerd who received a Ph.D. from Princeton University, Amodei dreamed of becoming a scientist. “I grew up in San Francisco in the whole tech boom. Google, Yahoo, Microsoft, they were all happening," he told The Wall Street Journal. “I had zero interest in it. None. Zero." He first saw the promise of AI as a tool to enable better healthcare research as a postdoc at Stanford University, he said at a company event in Washington Monday.

Amodei joined OpenAI shortly after it was founded as a nonprofit, then left in 2020 after clashing with Altman, its chief executive, over safety to start Anthropic. He is a believer in the earn-to-give movement, and committed to donating 80% of his founding stock to charity alongside his co-founders—a stake now worth billions of dollars.

A vegetarian since childhood, Amodei, now 42, often dotes on the chickens he keeps in a coop in his backyard, outfitted with a camera so he can watch over them. His Slack profile picture shows him smiling with a stuffed panda, and his office has a stuffed animal Amodei fondly calls “the wise octopus."

He is also the AI CEO most vocal about the technology’s potential to end civilization, warning that there is a 10% to 25% chance that AI goes rogue and unleashes planetary chaos. Around the start of Trump’s first term, Amodei warned in an AI presentation to industry colleagues that giving Trump control of powerful AI would be dangerous, and compared him in a slide to Xi Jinping and Vladimir Putin.

Amodei chose not to release an early version of Claude in the summer of 2022, fearing that it would start a dangerous technology race. Some Anthropic employees also indicated in a Slack poll they didn’t want to release the chatbot for the same reason. OpenAI released ChatGPT a few weeks later, forcing Anthropic to play catch-up. Amodei said he doesn’t regret the decision.

Sacks, 53, is Amodei’s almost perfect foil.

While attending Stanford University, he joined a conservative newspaper founded by Thiel, where he attacked political correctness on campus. The pair co-wrote a book called “The Diversity Myth" and worked together, along with Musk, in the early days of PayPal.

Sacks lives by Silicon Valley’s “move fast and break things" ethos. In 2012, after reports he would sell his startup to Microsoft for more than $1 billion, the South Africa native threw a Marie Antoinette-themed “Let Them Eat Cake" 40th birthday party, attended by Snoop Dogg and Charlie Sheen.

“Part of believing in capitalism is you don’t have to feel guilty about wanting to make money," Sacks told the New Yorker around that time.

While an executive at the HR startup Zenefits, Sacks handed out copies of Sun Tzu’s “The Art of War" and supported a plan to attack a rival company during a business dispute.

During the last election, Sacks was one of Trump’s most loyal supporters in Silicon Valley. He hosted a $12 million fundraiser and had Trump on his podcast. After being named AI czar, Sacks became one of the most powerful tech officials in Washington, and executives rushed to win his favor.

The exception was Amodei, who aligned with former President Joe Biden’s top tech officials. Amodei supported a 2023 executive order that put guardrails around the country’s best models and backed restrictions on chip exports to prevent countries like China from developing cutting-edge AI.

Policy pivot

Tensions began almost from the start of Trump’s second term.

While top tech executives attended the inauguration, Amodei attended the World Economic Forum in Davos. Anthropic then scooped up three top Biden-era officials. Sacks viewed the hires as a signal that the company had little interest in trying to support his agenda, people familiar with his thinking said.

Sacks was also suspicious of Anthropic’s ties to what he felt was a vast anti-Trump bureaucracy funded by Open Philanthropy, a nonprofit affiliated with the controversial effective altruism movement. The philosophy focuses on charitable giving and containing dangerous AI systems before they potentially destroy humanity, an approach that has divided Silicon Valley.

Open Philanthropy poured millions of dollars into think tanks and AI fellowships under the Biden administration, worrying Sacks and other conservatives. Anthropic took early funding from Open Philanthropy’s main backer, the Meta Platforms co-founder and Democratic donor Dustin Moskovitz, and its executives are close friends with—and in one instance married to—some of the group’s former leaders.

A spokesman for Open Philanthropy said the organization is nonpartisan and funds a wide range of projects.

One of Sacks’s first policy moves was to help roll back Biden-era restrictions and allow U.S. executives like Nvidia’s Jensen Huang to export more chips. He saw recent Chinese AI advances as proof Biden’s strategy didn’t work. In May, he traveled with Trump to the Middle East for the president’s first overseas trip, pledging to send chips to countries in the region. Tech executives including Altman joined the trip.

When he returned home, Sacks was surprised by the blowback to the agreements from some national-security hawks, said the people familiar with his thinking. He placed some of the blame on Anthropic and its allies.

“This is a national-security issue and not an economic issue," Amodei said at the Monday event. “Some of the elements in government don’t get it and are doing exactly the wrong things."

Inside the administration, some officials also questioned Sacks’s motives, given that the Middle Eastern countries he helped engineer deals for could back his portfolio companies or Craft Ventures, the venture firm where he still works. An official close to Sacks said two ethics waivers detailing Sacks’s investments show there isn’t a conflict.

Sacks soured further on Anthropic when he learned the company was lobbying against a proposed 10-year moratorium on state AI regulation, a bill supported by many tech companies.

Then Amodei publicly warned in late May that AI could destroy about half of all entry-level white-collar jobs, countering the administration’s message about AI benefiting the economy.

On his podcast, Sacks went on a 12-minute rant about Anthropic and its purported ties to effective altruism. He also criticized Amodei’s decision to add former Netflix CEO and Democratic donor Reed Hastings to Anthropic’s board.

At the end, Sacks said those in the network “needed to be Loomered," shorthand for getting attacked by the ring-wing activist Laura Loomer, who has publicly shamed those she sees as Trump’s opponents to get them fired. Earlier in the administration, she had called out members of the National Security Council, including tech-focused officials who opposed Sacks’s plans. They were then let go.

Shortly after the outburst, some of Amodei’s policy staff warned that speaking out against the moratorium could further antagonize the administration. Amodei went ahead, telling them it was a matter of principle and could influence future policy discussions. “To fully realize AI’s benefits, we need to find and fix the dangers before they find us," Amodei wrote in an opinion column in the New York Times in early June.

Around that time, a top AI policy official working with Sacks questioned Amodei about the move. “Why would you come out so strongly on this?" the official said.

The moratorium eventually fell apart after Steve Bannon and MAGA influencers fought to kill it over worries it would give tech companies too much power. “This was a populist revolt against the tech bros," Bannon said in an interview.

Woke AI

The administration ramped up work on AI executive orders in June. Much of the effort focused on making it easier to build data centers for training AI models around the world.

Sacks and other officials were also focused on woke AI. In early 2024, Google’s Gemini drew backlash when it changed the race or sex of some historical figures like George Washington in response to requests for images of those figures.

The officials drafted an executive order titled “Preventing Woke AI in the Federal Government." The order says the government “has the obligation not to procure models that sacrifice truthfulness and accuracy to ideological agendas." Anthropic wasn’t explicitly mentioned but was widely viewed by many in Washington as a target.

Some in the administration saw Sacks’s attacks against Anthropic as odd, given the quality of its models, and viewed the executive order as a potential effort to boost Musk’s AI startup xAI, which explicitly aims to be politically neutral. The official close to Sacks said the order had nothing to do with any one company. When the order was signed on July 23, Musk had already left the administration and had a public feud with Trump.

“Neither woke, or for that matter opposition to woke, has ever had anything to do with what Anthropic is aiming to accomplish in the world," Amodei said in the interview. He has said the company wants to work with the administration on areas where they agree, like increasing energy production and deploying AI safely in schools.

The order doesn’t apply to national security contracts—and just before Trump signed it, Anthropic announced a $200 million deal with the Department of War to forecast security threats. Shortly afterward, Anthropic said it would give government agencies access to Claude for $1, following a similar deal by OpenAI.

Yet according to Anthropic’s terms and conditions for all users, Claude can’t be used for any actions related to domestic surveillance. That effectively prevents many law-enforcement officials—including those at the Federal Bureau of Investigation, Secret Service and Immigration and Customs Enforcement—from using it.

Administration officials expected agencies would quickly receive waivers from Anthropic to use Claude in law enforcement, but they haven’t been granted, frustrating some in the administration. The OpenAI agreement has no such restrictions.

Meanwhile, investors were lining up to pour more cash into the company.

Amodei published a 1,300-word essay on the company’s Slack explaining his decision to seek funding from the United Arab Emirates and Qatar in mid-July, despite his concerns about “enriching dictators" and his opposition to building data centers in the region. Qatar’s sovereign-wealth fund recently invested in Anthropic’s $13 billion funding round.

After news of his essay leaked, Sacks shot back. “Anthropic criticized President Trump’s AI investments deals with the Middle East, but now it seeks investment from those very same countries," he wrote on X, formerly known as Twitter. “Perhaps it now realizes the absurdity of refusing to do business with the Gulf States."

Sacks was by Trump’s side a day later when the executive orders were signed at a summit co-hosted by his podcast. Trump praised Sacks at the start of a rollicking speech in which he bashed Biden’s approach and said people don’t want “woke Marxist lunacy" in AI models. During the event, tech executives lauded the administration’s strategy.

Loomer attacked Anthropic on X shortly afterward, sharing a story from a conservative media outlet about Anthropic’s Pentagon contract and noting Amodei’s pre-election Facebook post.

“We went into this fully knowing that there would be some people who would interpret policy disagreement as political opposition," Amodei said. “There is really nothing we can do about that."