The Messy Fight Over Attorney Fees in the Anthropic Settlement
A courtroom battle over legal fees is threatening to unravel what many hoped would be a clean resolution to one of the most closely watched copyright disputes in the AI industry. An attorney representing authors confirmed this week that copyright holders filed claims covering more than 92 percent of the roughly 480,000 works included in the anthropic copyright settlement. On the surface, that participation rate looks like a strong endorsement. But beneath those numbers, a fierce disagreement has erupted over how the money should be divided and whether the lawyers who negotiated the deal served their clients well or settled too quickly for their own benefit.
Objectors are pushing back hard, arguing that attorney compensation should be calculated based on the number of people filing claims rather than the total dollar value of the settlement fund. That distinction matters enormously, because shifting the formula could redirect millions of dollars from legal fees back to authors whose works were used to train AI models without permission.
The Core Dispute: Percentage of Fund Versus Per-Capita Fees
In most large class action settlements, attorneys request a percentage of the total fund as their fee. That approach means the lawyers take home a cut of the entire pot, regardless of how many people actually step forward to claim their share. In this case, the requested fee reportedly approaches one-third of a billion dollars. For individual authors, that sum feels outsized compared to what they stand to receive.
Objector Story has urged the court to make what he calls reasonable and fair adjustments to lower attorney fees. By shifting the calculation toward a per-claimant model, he argues, the money would flow more directly to the people whose creative work was used without authorization. Offering an example, he noted that a still-generous counsel payout of $70 million would yield a nearly 25 percent increase in individual plaintiff awards. Under that scenario, attorneys would still receive compensation equivalent to their current top hourly rates for the hours they actually worked, just not a windfall on top of it.
To Story, it also seemed like the attorneys could have pursued additional compensation for authors but chose not to. He claimed they settled far too quickly to maximize their own payout rather than exploring creative options that might have secured better terms for the writers they represented.
A Harsh Assessment of the Settlement Terms
Story did not hold back in his criticism. He stated plainly that if the attorneys were as skilled, gritty, and brilliant as they profess, and if the settlement were truly the home run that counsel claims it to be, plaintiffs would receive more than this pittance. His language reflects a deeper frustration that class action structures sometimes allow lawyer compensation to overshadow the relief provided to the actual victims of the alleged harm.
Ruben Lee, another class member who objected to the terms, agreed with that sentiment. He said the amount offered is paltry and does not in any way reflect the full value of the unauthorized use of his work. For authors who spent years developing their craft, receiving a fraction of what their work generated for a major AI company feels like an injustice rather than a resolution.
Why Attorney Fee Structures Matter in Class Actions
Many people outside the legal profession do not realize how fee calculations work in large class action settlements. Understanding that mechanism is essential for anyone affected by the anthropic copyright settlement or similar disputes involving AI training data.
In a typical class action, the attorneys who file the lawsuit negotiate a settlement with the defendant. That settlement creates a common fund from which both the plaintiffs and the lawyers are paid. The court then reviews the proposed fee and decides whether it is reasonable. Historically, courts have often approved fees that amount to roughly 25 to 33 percent of the total fund, especially in cases where the attorneys took significant risk by advancing costs upfront.
But critics argue that this percentage-based model creates a perverse incentive. If the goal is to maximize attorney compensation rather than plaintiff recovery, lawyers might accept a quick settlement that generates a large fee even if individual claimants receive relatively small amounts. Objectors in this case believe that exact dynamic played out.
By urging the court to base fees on the number of claimants rather than the fund size, they hope to align attorney incentives with plaintiff outcomes. That approach would reward efficiency and broad participation rather than sheer dollar volume.
What a Fair Fee Structure Would Actually Look Like
A fair fee structure in a case like this might blend two methods. First, the court could calculate a lodestar figure, which multiplies the hours worked by a reasonable hourly rate. Second, the court could apply a multiplier to account for risk and complexity, but cap that multiplier at a level that still leaves substantial money for the plaintiffs.
Story’s suggestion of $70 million in fees fits this blended model. At that level, attorneys would recover their full hourly value plus a premium for risk, but individual awards would increase by about 25 percent. That outcome strikes a balance between rewarding legal work and respecting the contributions of the authors who made the case possible in the first place.
For an author deciding whether to stay in the settlement or opt out, understanding this fee dispute is critical. If the court reduces attorney fees, the per-author payout goes up. If the court approves the higher fee, authors may feel shortchanged. That uncertainty creates a difficult decision point for anyone considering their options.
The Threat of Appeal and What It Means for the Settlement
Objectors have seemingly persuaded the court to at least entertain their strongly worded pleas. Among the most serious warnings is that the settlement may not survive an appeal if the terms are not reexamined. An appellate court could reject the entire deal if it finds that the fee arrangement unfairly prejudices class members.
That possibility adds significant pressure. If the settlement is overturned on appeal, everyone goes back to square one. Authors would have to wait longer for any compensation, and Anthropic would face continued legal uncertainty around its training data practices. A prolonged fight benefits neither side, which is why both parties have an incentive to address the objections now rather than risk a costly reversal later.
For a tech policy analyst watching this case, the appeal risk signals something larger. Courts are beginning to scrutinize AI copyright settlements more carefully than they might have a few years ago. The novelty of the legal questions, the asymmetry in bargaining power between individual authors and large AI companies, and the sheer scale of the works involved all make these cases unusually susceptible to challenge on appeal.
How Opting Out Creates Parallel Pressure
Notably, the objections came shortly before a group of 25 class members who opted out of the settlement filed a new lawsuit against Anthropic. That move shows that the company is not done fighting these claims even if the settlement is approved. Opt-out plaintiffs preserve their right to pursue individual or group litigation outside the class action framework.
For an author considering their options, opting out carries both risks and rewards. On the one hand, you forgo any guaranteed payout from the settlement. On the other hand, you retain the ability to negotiate directly or litigate for a larger recovery if your work has particularly high commercial value. The 25 authors who filed a new lawsuit clearly believe they can do better outside the settlement than inside it.
Story framed the broader stakes in stark terms. He said that for the court to agree that counsels request of nearly a third of a billion dollars while individual plaintiffs settle for a pittance of available compensation with no protections against future abuse is an aberration of civil justice and a slap in the face to all those who labored to publish their works. He added that such a decision would further the too often observed stereotype that class action plaintiffs are merely tools used to obtain Powerball-size payouts to attorneys.
Broader Implications for AI Copyright Law and Future Settlements
The anthropic copyright settlement is not happening in a vacuum. Several other major lawsuits are pending against AI companies that used copyrighted material to train their models without explicit permission. How this settlement is ultimately structured and approved will set a precedent for those cases.
If the court approves a fee arrangement that heavily favors attorneys, future defendants may see that as a license to offer relatively low plaintiff compensation knowing that the lawyers will still take a large cut. If the court instead rejects the current fee request and redirects money to authors, it sends a signal that future settlements must prioritize the people whose work was actually used.
You may also enjoy reading: Europe First to Authorize Moderna Combo mRNA Vaccine.
That precedent matters beyond this single case. The AI industry currently faces a patchwork of legal standards around training data. Some companies have struck licensing deals with publishers and content platforms. Others have relied on fair use arguments. The outcomes of these lawsuits will shape how companies approach data collection going forward.
What Authors Should Consider Right Now
If you are an author whose work may have been included in AI training datasets, this dispute has direct implications for your decision-making. Here are a few practical steps to consider.
First, review the settlement notice carefully. Look for the specific works listed and the compensation structure proposed. Compare what you would receive under the current terms with what you might expect from individual negotiation or litigation.
Second, monitor the fee objection hearings. If the court reduces attorney fees, the per-author payout could increase by a meaningful percentage. That change might make staying in the settlement more attractive than it initially appeared.
Third, consult with an attorney who specializes in copyright or class action litigation. A knowledgeable lawyer can help you evaluate whether opting out makes sense given the specific value and usage of your work. The 25 authors who filed a new lawsuit likely received advice that their works had above-average commercial potential.
Fourth, consider the timeline. Opting out typically has a deadline, and missing that window means you are bound by the settlement terms regardless of how the fee dispute resolves. Mark your calendar and gather your documentation early.
What Tech Companies Can Learn From This Dispute
For technology companies developing AI models, this case offers a cautionary lesson. Settling copyright claims quickly and quietly may seem efficient, but if the settlement terms appear unfair to the affected creators, the backlash can undermine the entire resolution. Engaging with authors and their representatives early in the process, rather than treating them as an afterthought, produces more durable outcomes.
Companies should also consider proactive licensing programs that compensate creators based on actual usage rather than forcing them into class action settlements years later. Several content licensing platforms have emerged that allow AI companies to pay for training data upfront. Those models reduce legal risk and build goodwill with the creative community.
What Comes Next in the Anthropic Case
The court now faces a difficult balancing act. Approving the settlement with the current fee structure risks an appeal that could kill the deal entirely. Rejecting the fee request and ordering a recalculation could satisfy objectors and strengthen the settlement against future challenges. But it also delays final approval and may antagonize the attorneys who spent years litigating the case.
Judge decisions in this area have varied widely across different districts and different types of class actions. Some courts treat percentage-of-fund fees as the default. Others apply a stricter lodestar cross-check and reduce fees that seem disproportionate to the work performed. The outcome in this case will depend heavily on how the judge weighs the objectors arguments against the attorneys track record.
For the broader landscape of AI copyright litigation, the next few months are pivotal. If the anthropic copyright settlement survives with adjusted fees, it could serve as a template for resolving similar disputes efficiently. If it collapses under the weight of fee objections and opt-out lawsuits, the industry may face years of fragmented litigation with no clear resolution.
Either way, the authors whose works powered some of the most advanced AI systems in existence deserve a fair outcome. The current fight over attorney fees is messy precisely because the stakes are so high. Getting the fee structure right is not just a technical legal question. It is a test of whether the class action system can deliver meaningful justice to creators in the age of artificial intelligence.
One thing is already clear. The days of treating copyright settlements in AI cases as routine approvals are over. Courts, objectors, and the public are paying close attention. And the attorneys who negotiated this deal may find that the most difficult part of the process is not the litigation itself, but the scrutiny of how they divided the proceeds.
For now, authors and their advocates continue to push for a fairer share. Whether they succeed will depend on the courts willingness to reexamine assumptions about fees that have gone unquestioned in many previous class actions. The outcome could reshape how every future AI copyright settlement gets built from the ground up.
