AI Usage Lawsuit
Anthropic Faces Federal Class-Action Lawsuit Alleging Misleading Claude Max Usage
Key Points
The complaint says Karl Kahn started on Claude Pro (June 2025) and then upgraded to Max 5x in January and Max 20x in April, alleging he “overpaid,” received less usage than advertised, and sometimes purchased additional usage after hitting plan limits.
The filing challenges not just the existence of caps but their specific structure: usage allegedly resets on “five-hour rolling windows” and then a separate weekly limit caps total activity, and Kahn argues neither constraint was clear enough to monitor.
Kahn’s lawyer Kati Daffan framed the case as a straightforward consumer-protection matter, telling Decrypt that “companies have to be honest about their advertising and marketing,” and that misrepresentations can lead to liability and refunds.
One report adds that Anthropic’s later weekly “power user” caps were implemented after heavy “Claude Code” use, and Anthropic said the change would “touch fewer than 5% of subscribers.”
That same report claims an enterprise customer ran up a “$500 million” bill in a single month—an example used to illustrate how quickly intensive AI use can translate into large compute costs under fixed-price subscription models.
Anthropic is facing a federal class-action lawsuit over claims that its priciest Claude AI subscription plans deliver far less than advertised. The suit, filed June 14 in the U.S. District Court for the Northern District of California, was brought by Washington, D.C. resident Karl Kahn, who paid up to $200 a month for a plan marketed as giving 20 times the usage of the base tier — and says he didn't get close to that CNET.
Kahn's lawyer, Kati Daffan, called it a simple consumer-protection case. "Companies have to be honest about their advertising and marketing," she told Decrypt. The proposed class covers all U.S. subscribers who bought a Max tier since April 9, 2025 — when Anthropic launched the $100 Max 5x and $200 Max 20x plans.
One Five-Hour Session Ate 15% of Weekly Allowance
Kahn started on Claude Pro ($20/month) in June 2025, upgraded to Max 5x ($100/month) in January 2026, and then moved to Max 20x ($200/month) in April 2026 to support intensive coding work with Claude Code, a terminal-based tool Engadget. During that top-tier plan, a single five-hour work session consumed roughly 15% of his entire weekly usage allotment, according to Quartz.
The complaint says two separate caps are layered on top of each other. First, a rolling five-hour window limits short bursts of activity. Second, a weekly ceiling caps total use. Kahn argues neither limit was clearly disclosed or easy to track, forcing him to sometimes buy extra access after hitting the wall Android Headlines.
Claude Code Power Users Triggered Tighter Caps Industry-Wide
The caps got stricter after heavy use of Claude Code pushed compute costs to unsustainable levels. Anthropic added weekly "power user" limits in response, saying the change would "touch fewer than 5% of subscribers" Tweaktown. But the timing stung users who had just paid premium prices expecting more headroom, not less.
The scale of the problem shows up in enterprise numbers too. One unnamed company reportedly ran up a $500 million bill in a single month from unrestricted Claude access, according to Axios. Internal Microsoft costs for Claude Code were estimated between $500 and $2,000 per engineer per month before limits were imposed. Those figures help explain why Anthropic moved to tighter controls — but they don't excuse unclear marketing, the lawsuit argues.
Why '5x' and '20x' Are Hard Promises to Keep
AI usage doesn't scale evenly. A short chat message costs a fraction of what a long coding prompt costs. When a user pastes a large file or generates hundreds of lines of code, the model processes far more "tokens" — the units that drive compute cost. That makes a "20x usage" promise almost impossible to deliver consistently across different types of work Tron Weekly.
Anthropic has also offered "2x usage" promotional bonuses in December 2025 and March 2026 to calm frustrated subscribers, suggesting the company was aware of the friction long before the suit was filed, according to Finout data cited in industry reports. On June 15 — one day after the lawsuit dropped — Anthropic rolled out a new billing structure that separates interactive chat from programmatic use and introduces metered credit pools, a shift away from the multiplier model entirely.
What the Lawsuit Could Force AI Companies to Change
If the case succeeds, Anthropic and rivals like OpenAI and Google may have to build real-time usage dashboards — a "digital fuel gauge" showing exactly how many tokens or compute credits remain in a billing cycle Beeble. Right now, subscribers have no easy way to verify whether they received 5x or 20x anything. The complaint asks for refunds and damages for the entire proposed class.
Anthropic is valued at $380 billion after a recent funding round and is weighing an IPO as early as October 2026 CloudZero. A consumer-protection ruling requiring clearer disclosures — or ordering refunds — could reshape how every major AI provider sells subscription plans. Some developers are already shifting toward open-source tools like DeepSeek, citing lower costs and more predictable usage, according to Gizmodo.
Sentiment
Mostly NegativePublishers
10
Articles
5
Reach
15
