From a tense corporate split to a viral photo: A timeline of Anthropic and OpenAI's budding rivalry

In July 2015, Tesla CEO Elon Musk, Altman, and a group of elite AI researchers all gathered at the swanky Rosewood Hotel in Menlo Park, California.

According to The New York Times, Musk had a falling out with then-Google CEO Larry Page. Weeks later, Musk, Altman, Amodei, Greg Brockman, Ilya Sutskever, and others discussed the formation of a new AI lab to ensure Google had a worthy competitor in the AI space. Musk invited Amodei, per tech journalist Alex Kantrowitz.

Their vision became OpenAI, though Amodei initially elected not to join the startup research lab. Roughly a year later, he changed his mind and joined OpenAI as "Team Lead for AI Safety."

Amodei quickly moved up the ranks at OpenAI. In September 2018, the startup named him its research director.

Altman, who cofounded OpenAI while still serving as president of Y Combinator, began to devote more time to the startup.

In March 2019, Altman stepped down as YC's leader. He then became CEO of OpenAI and led the startup's pivot to a capped for-profit structure.

In November 2019, OpenAI released GPT-2, which Amodei played a major role in developing. A month later, OpenAI named Amodei as its Vice President of Research.

In June 2020, OpenAI began to show just how far the technology had come with the release of GPT-3, considered to be the first highly capable Large Language Model (LLM).

Amodei told The New York Times that the model had "this emergent quality." Independent researchers told the publication that GPT-3's capabilities surprised them, even as the model still showed signs of struggle.

To address safety concerns, OpenAI initially controlled access through a private beta.

The release of GPT-3 solidified OpenAI's standing, but behind the scenes, tensions were rising.

The rifts began when Amodei successfully lobbied to keep Greg Brockman, an OpenAI cofounder, off the team that developed GPT-3, according to Keach Hagey's biography of Altman, "The Optimist: Sam Altman, OpenAI, and the Race to Invent the Future."

Hagey wrote that Amodei's stunning power within OpenAI had started to ruffle feathers. Differences continued to escalate over Amodei's long-held views on safety, Hagey wrote, especially regarding slowing the pace of updates to prevent malicious uses of the AI models.

Amodei told friends that he "felt psychologically abused by Altman," Hagey wrote. Altman, in turn, was telling colleagues that the tension "was making him hate his job."

On December 29, 2020, OpenAI made it official. Amodei was leaving, and a "handful" of other colleagues were leaving.

Amodei has since suggested that his vision became incompatible with OpenAI's direction.

"It is incredibly unproductive to try and argue with someone else's vision," Amodei told podcaster Lex Friedman in 2024, when asked why he left OpenAI.

With seven other former OpenAI employees, Amodei founded Anthropic in early 2021. The group was extremely close and included Daniela Amodei, Dario's sister. Daniela Amodei later said the name was chosen to emphasize their company's focus on humans.

Only one of Anthropic's initial employees hadn't worked at OpenAI, according to AI Business.

Despite starting from scratch, Amodei said that by the Summer of 2022, the company's chatbot, Claude, had finished training. Amodei said he was worried about what the release of a powerful AI could mean. Anthropic held off on a release.

"I suspect it was the right thing to do," Amodei told Time Magazine in 2024. "But it's not totally clear-cut."

Months later, OpenAI released ChatGPT, kicking off the AI race and making Amodei's former employer a household name.

As Anthropic began to establish itself in its own right, Amodei began to use his public appearances to take what were widely viewed as implicit shots at OpenAI.

During an appearance at a Bloomberg event, Amodei noted how Anthropic had kept its leadership intact.

"We have 7 cofounders," he said, Gizmodo reported. "Three and a half years later, we're all still at the company."

While never calling out by name, OpenAI was experiencing upheaval at the time. Months earlier, Andrej Karpathy, an OpenAI cofounder, had left the company. And in November 2023, Altman was briefly pushed out of OpenAI, an effort fellow cofounder Ilya Sutskever assisted. (Sutskever later expressed regret over his role. He formally left OpenAI just days after Amodei's jab, though there had been months of speculation surrounding Sutskever's standing.)

By the end of 2025, OpenAI's lead in the AI race was slipping. When presented with the opportunity, Amodei seized the moment to troll his rival.

Journalist Andrew Ross Sorkin asked Amodei about OpenAI's decision to declare a "code red" to marshal resources for ChatGPT amid Google's rising strength.

We have a little bit of a privileged position where we can just keep growing and just keep developing our models, and we don't have to do any code reds," Amodei told Sorkin during an appearance at The New York Times' DealBook summit.

Earlier in their conversation, Amodei appeared to take another swipe at Altman when he talked about some players "who are YOLOing" by making too risky bets on future demand based on their current revenue.

"Who is YOLOing?" Sorkin asked.

"I'm not going to answer that," Amodei replied.

Anthropic used the biggest stage available to take its most direct shot yet at OpenAI.

Ahead of the Super Bowl, Anthropic revealed it was spending millions on an advertising campaign to denounce AI chatbot ads. While OpenAI was not named directly, it was clear who the intended target was, given that just months earlier, OpenAI had said it would begin testing ads on ChatGPT.

The ads featured real human actors mimicking the voice of product-pushing AI chatbots when asked questions like how to get a six-pack quickly or how to better connect with your mom.

"First, the good part of the Anthropic ads: they are funny, and I laughed," Altman wrote in a lengthy post on X." But I wonder why Anthropic would go for something so clearly dishonest."

Altman wasn't done.

"Anthropic serves an expensive product to rich people," he continued. "We are glad they do that and we are doing that too, but we also feel strongly that we need to bring AI to billions of people who can't pay for subscriptions."

Days later, the companies went head-to-head again. This time, with the release of major updates to their coding-focused model within minutes of each other.

A who's who of AI and tech elite gathered in India for a major summit on artificial intelligence.

Indian Prime Minister Narendra Modi took the opportunity to orchestrate a classic image of unity: competing CEOs with their hands raised together. (It's something Modi has done before with other world leaders, and politicians have been doing forever.)

Modi almost got his moment. While Altman held the prime minister's hand, the OpenAI CEO didn't grasp Amodei's hand, who was positioned to his other side. Amodei grasped the hand of the other person next to him, but not Altman's.

The internet, predictably, had a field day. And the world got a perfect encapsulation of one of AI's bitter rivalries.