In early 2021, former employees of OpenAI started a new AI company: Anthropic. The driving force behind the venture was siblings Daniela and Dario Amodei. The ambition was to create an AI company that combined technical brilliance with security thinking and social impact, as it was called when Daniela Amodei herself tweeted about the venture.
Like many other AI companies, it took a couple of years for Anthropic to launch a product to the general public. After several successful funding rounds, ChatGPT competitor Claude was released in July 2023.
Basic principles
At first glance, the AI tool looked like a copy of ChatGPT. Simply a chatbot that seemed intelligent. The difference was behind the scenes.
From the start, Anthropic has said that Claude has, in somewhat simplified form, ethical and moral foundations. These foundations, built into how Claude works, have been expanded since the first version came out. Back then, in 2023, it consisted of 2,700 words - the latest version consists of over 23,000 words.
Over time, Claude has also become extremely good at writing code, something that has driven the value of the service and, by extension, Anthropic skyward.
Stock market impact
However, specialized parts of Claude have also created stock market chaos. In February, various launches of Anthropic linked to Claude caused billions of dollars in stock market value to disappear, Axios reports.
On Monday, February 23 alone, IBM's stock plunged by over 13 percent, wiping out $31 billion, according to Axios. The reason? Claude is said to be able to fix code in banking and business transactions - an important business area for IBM.
Choosing a path
In September 2025, Anthropic was valued at 1.6 trillion kronor. That valuation doubled by February this year. At the same time, clouds of worry are growing on the horizon. In January this year, the Washington Post revealed that the company had purchased millions of books since 2024 in order to scan the content and feed the information into Claude.
Recently, Anthropic accused Chinese competitor Deepseek of trying to copy and steal Claude's knowledge on an "industrial scale" using fake accounts.
But the big question is whether Anthropic will violate its own established rules. The company currently has a unique partnership with the Pentagon as the only AI company partner. However, the U.S. Department of Defense wants full access to Claude's capabilities without those established rules. Anthropic has until today, Friday, to respond.
