On March 14, 2023, Anthropic introduced Claude, describing it in its own announcement as “a next-generation AI assistant based on Anthropic’s research into training helpful, honest, and harmless AI systems.” Claude was made available through a chat interface and an API.
The announcement lists tasks Claude can handle, including “summarization, search, creative and collaborative writing, Q&A, coding, and more.” Anthropic emphasized three distinguishing qualities: that Claude is “much less likely to produce harmful outputs,” that it is “easier to converse with, and more steerable,” and that users can direct its “personality, tone, and behavior.”
Anthropic offered two versions at launch: Claude, the higher-performance model, and Claude Instant, a lighter, faster, and less expensive option.
Claude mattered as the arrival of a serious, safety-focused competitor to OpenAI’s offerings, founded by former OpenAI researchers and centered on a research agenda for aligning AI systems with human intent.