Anthropic, PBC is an American artificial intelligence (AI) public benefit corporation headquartered in San Francisco, California. Its flagship product is Claude, a series of proprietary large language models (LLMs). Anthropic was founded in 2021 with the goal of promoting AI safety by former members of OpenAI, including siblings Daniela Amodei and Dario Amodei, who are president and CEO, respectively. The company is privately held but reportedly plans an initial public offering in 2026. Valued at US$965 billion in a May 2026 Series H funding round, Anthropic is one of the world's most valuable AI pure play companies, rivaled by OpenAI.
Anthropic conducts research in areas such as mechanistic interpretability, safety, alignment, and societal impact. Notable researchers working at Anthropic include Andrej Karpathy, John Jumper, Jan Leike, Chris Olah, and Amanda Askell. The company's approach emphasizes careful testing and gradual deployment of AI systems, often contrasting with faster release cycles in the industry.
History
Anthropic was founded in January 2021 by seven former employees of OpenAI, including siblings Daniela Amodei and Dario Amodei, the latter of whom was OpenAI's Vice President of Research. They raised $124 million in May 2021. In the summer of 2022, Anthropic finished training the first version of Claude but did not immediately release it, citing a need for further internal safety testing and a desire to avoid initiating a potentially hazardous race to develop increasingly powerful AI systems. It was released in March 2023.
In 2024, Anthropic hired several notable employees from OpenAI, including Jan Leike and John Schulman. In May 2025, the company announced Claude 4. It also introduced new API capabilities, including the Model Context Protocol (MCP) connector. The company hosted its inaugural developer conference that month. Also in May, Anthropic launched a web search API that enables Claude to access real-time information from the internet. Claude Code, Anthropic's coding assistant, transitioned from research preview to general availability.
In July 2025, Anthropic signed a two-year contract with the US Department of Defense for $200 million. Anthropic LLMs became integrated with the Pentagon's classified intelligence networks and with Palantir's data platform. Anthropic stipulated that the Pentagon not use its technologies for mass surveillance of Americans, or to create fully autonomous weapons. Amodei clarified that Anthropic supports the use of AI in "lawful foreign intelligence and counterintelligence missions" and partially autonomous weapons, such as those used in the Russo-Ukrainian war.
In September 2025, Anthropic announced that it would stop selling its products to groups majority-owned by Chinese, Russian, Iranian, or North Korean entities due to national security concerns. In October 2025, Anthropic announced a cloud partnership with Google, giving it access to up to one million of Google's custom Tensor Processing Units (TPUs), potentially adding more than one gigawatt of AI compute capacity by 2026. In November, Nvidia and Microsoft were expected to invest up to $15 billion in Anthropic, and Anthropic said it would buy $30 billion of computing capacity from Microsoft Azure running on Nvidia AI systems.
In November 2025, Anthropic said that hackers sponsored by the Chinese government used Claude to perform automated cyberattacks against around 30 global organizations. The hackers bypassed Anthropic's safeguards by pretending it was for defensive testing. In December 2025, Anthropic acquired Bun to improve Claude Code's speed and stability. The same month, Anthropic signed a multi-year, $200 million partnership with Snowflake Inc. to make Claude models available through that company's platform.
Claude was reportedly used during the 2026 United States intervention in Venezuela. In February 2026, Anthropic aired two commercials during Super Bowl LX as part of a broader marketing campaign called "A Time and a Place", with four ads created by Mother. Each ad depicts AI assistants suddenly pivoting to promoting a fictional product in the middle of a conversation. Anthropic said that Claude will stay ad-free, in contrast to its competitor OpenAI, which introduced ads to the free version of ChatGPT. The same month, Anthropic donated $20 million to Public First Action, an AI regulation advocacy group.
On February 23, 2026, Anthropic accused Chinese competitors DeepSeek, Moonshot AI, and MiniMax Group of knowledge distillation "attacks" in violation of Anthropic's terms of service. Anthropic alleged the companies in total generated over 16 million chats with Claude using around 24,000 fraudulent accounts, which were then used as training data. In June 2026, Anthropic accused Alibaba Cloud of similar practices.
In 2026, Anthropic refused the United States Department of Defense's (DoD) demand to remove contractual restrictions prohibiting the use of its AI technology for surveillance and autonomous weapons. After the refusal, the DoD designated the company a "supply chain risk" and barred all U.S. military private contractors, suppliers, and partners from doing business with the firm (see Anthropic–United States Department of Defense dispute). At the outset of the 2026 Iran war, the US military used Claude to identify and rank targets in Iran according to their strategic importance, to generate strike sequences, and to write legal justifications for the strikes.
In May 2026, Anthropic announced a deal with xAI for use of its Colossus 1 data center to expand its model capacity. On May 18, the company acquired software startup Stainless for an undisclosed sum. It also made
Products and Technology
Anthropic develops the LLM Claude in four tiers: Haiku, Sonnet, Opus, and Fable. Claude is accessible via online chatbot, API, and Anthropic's agent harnesses: Claude Code, Cowork, Design, and Science. Its most powerful model class, Claude Mythos, is restricted to partnered US organizations, with a cybersecurity and life sciences focus. Anthropic's Project Panama, internally described as an "effort to destructively scan all the books in the world", provides training data for Claude. Before Project Panama, Anthropic downloaded pirated books and was sued, settling in 2025 a copyright infringement lawsuit from authors for $1.5 billion.
Anthropic's research includes work on mechanistic interpretability, which aims to understand how neural networks process information. The company also explores alignment techniques such as RLHF and constitutional AI, though specific details are often kept confidential. Claude models are built on transformer architectures, similar to other LLMs, but with a focus on safety and controllability.
Safety and Alignment
Anthropic's core mission is to ensure that AI systems are safe and aligned with human values. The company emphasizes "constitutional AI," a method where models are trained to follow a set of principles, reducing reliance on human feedback. This approach is designed to make AI more transparent and less prone to harmful behaviors. Anthropic also conducts red-teaming and stress-testing of its models before release, as seen in the delayed launch of the first Claude.
The company has been involved in public policy discussions, advocating for responsible AI development. In February 2026, Anthropic donated $20 million to Public First Action, an AI regulation advocacy group. Anthropic's stance on military use is nuanced: it supports lawful foreign intelligence and partially autonomous weapons but draws the line at mass surveillance and fully autonomous weapons, a position that led to the dispute with the DoD.
Partnerships and Controversies
Anthropic partners with Palantir to provide Claude to US federal agencies including the Department of Defense (DoD) and the Intelligence Community. Claude was used in the 2026 Iran war and the 2026 United States intervention in Venezuela. In February 2026, the DoD demanded the ability to use Claude for all lawful purposes, but Anthropic refused to drop restrictions against mass domestic surveillance and fully autonomous weapons. The dispute led the Trump administration to phase out Anthropic products from government use, a decision blocked by a preliminary injunction from a federal judge who called it "First Amendment retaliation".
The company has also faced accusations of intellectual property issues. In February and June 2026, Anthropic accused Chinese competitors DeepSeek, Moonshot AI, MiniMax Group, and Alibaba Cloud of distilling Claude for their models. These allegations highlight the competitive tensions in the AI industry, especially regarding model training data and proprietary techniques.
Future Outlook
As of 2026, Anthropic is positioned as a major player in the AI landscape, with a valuation of $965 billion and plans for an IPO. The company continues to expand its capabilities, including partnerships with cloud providers like Google and Microsoft Azure, and acquisitions such as Bun and Stainless. However, its commitment to safety and ethical constraints may limit its adoption in certain sectors, particularly military applications. The ongoing legal and political battles with the US government could shape its future trajectory, but Anthropic remains a key voice in the debate over AI safety and governance.