Altman and Amodei: How OpenAI Co-Creators Became Bitter Rivals

A moment at an AI summit in New Delhi last month captured the deep rift between two of artificial intelligence's most powerful figures.

When Indian Prime Minister Narendra Modi raised the hands of AI leaders including Google CEO Sundar Pichai for a group photo on February 19, cameras focused on Sam Altman of OpenAI and Dario Amodei of Anthropic. While other participants smiled and clasped hands, the two CEOs refused to shake hands, standing with awkward expressions and forced smiles.

The image went viral on social media. Siddharth Bhatia, co-founder of AI startup PuchiAI, posted on X: "When will AGI arrive? The day Dario and Sam hold hands." Justin Moore, an investment partner at Andreessen Horowitz, shared the photo with the caption: "When you're forced to do a group project with your boss."

The two were once colleagues at OpenAI. Amodei joined from Google Brain as a senior researcher and led development of GPT-2 and GPT-3, laying the foundation for ChatGPT. He served as vice president of research while his sister Daniela Amodei held the vice president role for safety and policy.

The relationship fractured when OpenAI sought Microsoft investment and began transitioning from nonprofit to for-profit status. Amodei, deeply concerned about AI ethics and safety, criticized the Microsoft deal as pushing OpenAI toward excessive commercialization.

The Amodei siblings departed in 2020 and founded Anthropic with former colleagues, aiming to build AI under stricter controls. The company has since raised billions from Google, Salesforce, and Amazon, and its Claude chatbot now directly challenges ChatGPT.

"After developing GPT-2 and GPT-3 at OpenAI, I held two strong convictions," Amodei told Fortune in 2023. "First, that adding more compute would continuously improve model performance with no ceiling. Second, that beyond scaling, we needed alignment and safety—because more compute alone cannot instill values in a model."

Competition intensified as both companies approach potential IPOs. Anthropic's valuation has reached approximately $380 billion, chasing OpenAI's $840 billion.

Tensions flared further when Anthropic aired a Super Bowl advertisement last month appearing to mock ChatGPT's responses. Altman called it "deceptive and dishonest."

The companies clashed again over Trump administration AI policy. When the Defense Department pressured Anthropic to allow unrestricted military use of Claude, the company refused, citing opposition to mass surveillance of citizens and fully autonomous weapons. President Trump denounced Anthropic as "radical left" on February 27 and ordered federal agencies to stop using its technology.

OpenAI subsequently signed a Defense Department contract, but faced user backlash for appearing to capitulate where Anthropic resisted. Anthropic reported Claude's free users increased over 60% since January, with paid subscribers doubling. On February 28, Claude surpassed ChatGPT as the top free app on Apple's U.S. App Store.

Facing criticism, Altman announced on March 2 that OpenAI is adding contract provisions prohibiting intentional use of AI for domestic surveillance of U.S. residents or citizens. "We shouldn't have rushed the announcement on February 27," he wrote. "This issue is very complex and required clear communication. As a result, I think we looked opportunistic and sloppy."