AI power users claim Anthropic duped them with subscriptions, and they’re taking it to court

The Verge reports an expanded class action filed by attorneys Monica Vaca and Kati Daffan—both formerly of the FTC under Lina Khan—alleging Anthropic deceptively advertised Max tier limits. Max upgrades from the $20/month Pro plan to $100/month for “5x” Pro usage or $200/month for “20x,” but the complaint says those multipliers apply inside five-hour chunks that are also subject to a weekly limit, yielding far less total capacity than marketing graphics imply.

Plaintiffs say understanding the real terms required clicking through multiple hyperlinks to find “session,” then leaving to the Pro page for what “session” means. Power users often upgrade mid-project when they hit limits, then discover the bump is not what the pricing page suggested. A Reddit analogy quoted in the piece: a bigger gas tank with a pump still limited to one gallon every five hours.

Anthropic announced Max in April 2025 and, per the suit, imposed the contested weekly limits months later in August while competing with OpenAI. An earlier complaint filed in July was withdrawn and refiled expanded; Anthropic’s motion to dismiss argued clarifying info was a click away—“the digital equivalent of flipping a product over to read the back label.” Anthropic did not comment to The Verge.

Signal Desk’s read: this case is less about one lab’s fine print and more about whether AI subscriptions can keep selling multiplier metaphors while metering reality through opaque session and weekly gates.

Anthropic has publicly prioritized power users—even when that meant cutting off popular applications like OpenClaw—yet those same customers are the ones alleging bait-and-switch. That contradiction matters. If Max is the loyalty product for heavy Claude use, trust in the meter is the product. Marketing “5x” and “20x” in graphics while burying session definitions two links deep is exactly the pattern false-advertising doctrine was built to police, which is why ex-FTC counsel makes this filing more than Twitter grievance.

Anthropic’s defense—information was available if you hunted—may be literally true and still lose the plot. Vaca’s point that consumers cannot audit a black-box usage system and must rely on marketer claims is the stronger frame for a $100–$200/month tool people feel career pressure to buy. The August addition of weekly limits after an April launch, if proven as alleged, also supports a bait-then-constrain narrative rather than a stable disclosed plan.

What is under-stated industry-wide is how cost pressure is being pushed into UX: Fable 5.1’s own nod to customer price feedback sits in the same climate. What is over-stated is that every usage complaint is fraud; metering expensive inference is rational. The legal question is disclosure honesty, not whether limits exist. Expect peer labs’ pricing pages to get lawyered next—multiplier language without plain-language total weekly capacity is now litigation bait.

AI labs have spent months raising prices and tightening limits as they chase margins. Anthropic’s Max plan sits above Pro as the power-user tier; the suit sits in a broader consumer-protection turn aimed at AI commercial speech, not model safety.