Anthropic

Anthropic’s product story is Claude: capable, controllable, and positioned for regulated workflows. Its policy story is reliability under distribution—how guardrails and evals behave when models are embedded in real software stacks, not demo chat UIs.

We treat Anthropic like any other lab: strong engineering can coexist with overstated macro claims. The company hub links models, claims with named leadership quotes, and post-release evidence about behavior shifts—not vibes about “helpfulness.”

Tracked claims attributed

• Everybody got super excited last year, but we are running out of improvements; at least for a little… [refuted]
• In 2023 and early 2024, some high-performing organizations were already reporting measurable EBIT im… [open]
• AI will affect almost 40 percent of jobs around the world, replacing some and complementing others. [open]
• cancer is cured, the economy grows 10% a year, the budget is balanced—and 20% of people don't have j… [refuted]

Models shipped

• Claude 3.5 Sonnet — *Anthropic mid-2024 workhorse tuned for coding and long-context tasks at competitive latency.*
• Claude Opus 4 (x) — *Anthropic flagship-oriented Opus line refresh pitched for highest-stakes analysis, coding, and agentic workflows.*