Anthropic is facing a proposed class-action lawsuit over how it sells its two most expensive Claude plans. The case, filed June 14 in a California federal court, is straightforward: a paying customer says he was charged $200 a month for a level of access that was advertised, but never actually delivered.
The plaintiff, Karl Kahn, started on Claude's base paid plan and gradually moved up. He eventually upgraded to the Max 20x tier in April, after finding the lower plans too restrictive for his coding work. One five-hour session used up 15 percent of his weekly allowance. Within days, the limits were cutting off his work entirely, leaving him to either stop, slow down, or pay again on top of his subscription.
The core of the lawsuit is not that limits exist. Weekly caps on heavy usage are real, and Anthropic introduced them in 2025 specifically to stop certain users from running Claude continuously around the clock. The lawsuit's argument is that these limits were sold dishonestly: the Max plans advertise five and twenty times the usage of the base Pro plan, but no one can tell what that base figure actually is, because it shifts depending on demand. That makes the promised multiplier essentially unverifiable.
The complaint also points to emails Anthropic allegedly sent in July 2025 that outlined, for the first time, what weekly usage subscribers at each tier should actually expect per model. The fact that this clarification came after the plans launched, and after users started complaining, is central to the case.
For anyone paying $100 or $200 a month for a professional AI tool, this situation points to something worth understanding. AI services do not work like streaming subscriptions. On Netflix, one more viewer costs the company almost nothing. On Claude, one heavy user running intensive coding sessions can consume hundreds of dollars worth of computing resources in a matter of hours. That is why AI companies set usage caps in the first place, and why those caps tend to be vague: the actual cost of serving any given user is unpredictable.
The honest version of an AI subscription would tell you exactly how much you can use and in what units, before you pay. The industry, so far, has largely avoided that level of clarity. Anthropic is not alone in this. Most AI subscription pages describe limits in relative terms, ratios and multipliers anchored to a baseline that shifts with server demand.
This lawsuit is a signal that those days may be numbered. Consumer protection law, as the plaintiff's legal team notes, requires companies to be honest about what they are selling. A plan called "20x" implies a specific and reliable amount. Courts may well agree.
If you are a business currently paying for premium AI subscriptions, or evaluating whether to, the practical takeaway is simple: ask the vendor directly what your monthly limit is, in concrete terms, before you commit. If they cannot answer that question clearly, that itself tells you something about how reliable the service is for professional use.