Anthropic: The Company Behind Claude

Who Is Anthropic?

Anthropic is an AI safety research company founded in 2021 by Dario Amodei (CEO) and Daniela Amodei (President), along with several colleagues from OpenAI.

They left OpenAI over disagreements about how quickly to scale AI systems and how much attention was being paid to safety research. Anthropic was built around the premise that AI safety needs to be the primary focus, not an afterthought.

Headquartered in San Francisco. About 1,000 employees as of 2025.

The Money Behind Anthropic

Anthropic has raised over $7.3 billion in funding. The two largest investors:

• Amazon — committed up to $4 billion. AWS (Amazon Web Services) runs Anthropic's cloud infrastructure. Claude models are available through Amazon Bedrock.
• Google — invested approximately $2 billion. Anthropic models are available through Google Cloud's Vertex AI.

Both Amazon and Google are also OpenAI competitors through their own AI products. Investing in Anthropic gives them access to a second AI capability track alongside their internal research.

Constitutional AI — How They Train Claude

Anthropic developed a training approach they call Constitutional AI. The idea: instead of relying only on human feedback to teach the model what responses are good, they give the model a set of principles (a "constitution") and have it evaluate its own responses against those principles.

The constitution includes things like: be helpful, be honest, avoid harmful outputs, avoid deception. The model critiques its own drafts and rewrites them to better fit these principles.

This approach gives Anthropic more systematic control over the model's behaviour than traditional reinforcement learning from human feedback alone. It's also more auditable — you can read the principles and see what Claude was trained to prioritize.

Claude — Their Product

Claude is Anthropic's AI assistant, available at claude.ai.

• Claude Free — limited daily messages, Claude 3.5 Sonnet model
• Claude Pro — ~$28 CAD/month, 5× more usage, Claude 3.5 Opus, Projects feature
• Claude API — for developers, pay per token used

Claude is generally preferred over ChatGPT for writing long documents, nuanced editing, and tasks that require the AI to follow complex instructions precisely. Both are capable — the preference often comes down to tone. Claude tends to be more direct and less verbose.

Anthropic and Canadian Users

Claude is fully available in Canada. English and French are both supported.

Privacy note: conversations are processed on US servers. Anthropic states they do not sell your data and do not use it to train models by default (unlike OpenAI, which trains on user data unless you opt out). You can verify this in their privacy policy.

PIPEDA applies to Canadian users' data. If you work with sensitive Canadian professional data (medical, legal, financial), check your organization's data handling policies before pasting it into any AI tool.

Anthropic vs OpenAI

Both companies are making powerful AI. The differences are mostly philosophical and structural.

OpenAI is larger, has more products (ChatGPT, Sora, DALL-E, Operator), and has 200M+ weekly users. Their model has moved toward commercial scale fast.

Anthropic publishes more safety research. Their Constitutional AI work, their research on model interpretability (trying to understand what's happening inside the model), and their evaluations of dangerous capability emergence are all public.

For most users, the practical difference is: Claude is better for writing and long-context tasks, ChatGPT has more features and integrations.