- Let’s take a look at the Susman Godfrey strategy. Today, I will discuss 4 strategies:
- The Shadow Library Strategy
- The Class Action Strategy That Includes Statutory Damages
- The Anti-RAG Strategy
- The Antitrust Strategy
No law firm has had a greater impact among the 50 U.S. copyright lawsuits against AI companies than Susman Godfrey.
Among lawyers, the firm needs no introduction. In fiscal 2023, it ranked No. 4 among law firms for profits per partner, with a hefty $7 million per partner.
While the Joseph Saveri Law Firm started the copyright AI lawsuit frenzy, filing the first of many lawsuits starting in January 2023, Susman Godfrey has developed a litigation strategy — what I call a playbook — that has had partial success in Bartz v. Anthropic, resulting in the proposed $1.5 billion settlement, the largest in U.S. copyright litigation ever.
Let’s take a look at the Susman Godfrey Playbook. Today, I will discuss 4 strategies (leaving others for future analysis):
- The Shadow Library Strategy
- The Class Action Strategy That Includes Statutory Damages
- The Anti-RAG Strategy
- The Antitrust Strategy
*Before we do, one major caveat: Susman Godrey isn’t the only law firm involved in at least some of the lawsuits I will discuss. So credit should also go to the other law firms involved in these lawsuits. It’s too hard from the public record for me to apportion responsibility, so, for simplicity, I will refer to Susman as the lead in these cases. (In Bartz, Lieff Cabraser and Cowan Debaets are also representing the plaintiff class.)
1. The downloading of pirated books from shadow libraries
The Shadow Library / Pirated Books Strategy: The Shadow Library Strategy is to raise a separate theory of infringement–apart from the training of AI models–based on the AI company’s initial acquisition of copies of works from controversial shadow libraries.
Interestingly, Bartz did not file for summary judgment on this issue. Instead, they opposed Anthropic’s motion for summary judgment on fair use. Given that Judge Alsup all but ruled in favor of Bartz on the initial acquisition and library building from shadow libraries, we can expect that Susman Godfrey will move for summary judgment in the other copyright lawsuits it is involved in. Here’s the argument by Susman in its opposition in Bartz v. Anthropic:
Of course, the downloading of copies from shadow libraries was not a strategy unique to Susman Godfrey. The Boies Law Firm was pursuing the exact same strategy in Kadrey v. Meta, but Judge Chhabria rejected the argument of separate use and instead viewed the initial acquisition from shadow libraries by Meta as for the “further purpose” of training its AI models.
Judge Alsup’s Added Wrinkle: Acquisition For the Purpose of Permanent Library Building: Judge Alsup did not embrace the Bartz argument advanced by Susman entirely. Instead, Judge Alsup framed Anthropic’s separate use of downloading from the shadow libraries for the purpose of storing the copies in a permanent library (aka library building).
Anthropic said this theory of library building wasn’t even argued by Bartz in its opposition (true), so perhaps not all the credit should go to Susman on the theory that ultimately prevailed.
In any event, what we do know is that the Shadow Library Strategy is proliferating in other copyright lawsuits against AI companies. In its separate lawsuits against Anthropic, Concord Music, for example, has asked Judge Lee for leave to add a new claim of infringement based on the Shadow Library Strategy.
The recently filed lawsuits against Microsoft (filed by Susman) and Apple adopt this strategy as well.
The success of the Shadow Library Strategy hinges on whether the court treats the initial acquisition of copies as a separate use as Judge Alsup did (for the purpose of library building) or as a part of the same use for the further purpose of training the AI model as Judge Chhabria did in Kadrey.
In the book author lawsuits against OpenAI and Microsoft, the irony is that the same person — Benjamin Mann — was partly responsible for downloading copies from the shadow library LibGen in 2019 at OpenAI before he did so later at Anthropic, which he co-founded. Apparently, though, OpenAI destroyed some of the early books datasets it used, which might give it a way to distinguish the library-building rationale of Judge Alsup’s ruling.
The reason why the Shadow Library Strategy is so potent is that basically all the AI companies that developed LLMs are alleged to have downloaded some of the same books datasets because books proved to be higher quality data to train AI models and yielded better, state-of-the-art models, something that Judge Chhabria noted in his fair use opinion.
And the potential payoff in statutory damages is exactly the same, regardless of winning (or losing) on the issue of fair use in AI training. This under-appreciated fact illuminates the true potency of the Shadow Library Strategy. Prevailing on AI training would add $0 in statutory damages for plaintiffs.
2. The class action Strategy that includes statutory damages
The next strategy being advanced by Susman Godfrey and, indeed, many other plaintiffs’ firms in the 50 copyright lawsuits, especially the book author lawsuits, is seeking to certify a class action that includes the potential recovery of statutory damages under the Copyright Act (based on prior copyright registration of the works before the infringement started).
Thus far, Susman Godfrey in the Bartz case is the only one that has successfully obtained class certification. (No other court has yet ruled on such a motion yet.)
The reason why class certification that includes statutory damages is strategically important is that increases the potential recovery to astronomical levels. For basic infringement, here is the range of statutory damages:
| Works Infringed | $750/work | $1,000/ work | $10,000/ work | $20,000/ work | $30,000/ work |
| 10,000 | $7.5M | $10M | $100M | $200M | $300M |
| 100,000 | $75M | $100M | $1B | $2B | $3B |
| 250,000 | $187.5M | $250M | $2.5B | $5B | $7.5B |
| 500,000 | $375M | $500M | $5B | $10B | $15B |
| 750,000 | $562.5M | $750M | $7.5B | $15B | $22.5B |
| 1,000,000 | $750M | $1B | $10B | $20B | $30B |
| 1,500,000 | $1.125B | $1.5B | $15B | $30B | $45B |
| 2,000,000 | $1.5B | $2B | $20B | $40B | $60B |
| 3,000,000 | $2.25B | $3B | $30B | $60B | $90B |
| 4,000,000 | $3B | $4B | $40B | $80B | $120B |
| 5,000,000 | $3.75B | $5B | $50B | $100B | $150B |
| 6,000,000 | $4.5B | $6B | $60B | $120B | $180B |
If willful infringement is found, here is the range of statutory damages:
| Works infringed | $50K/work | $75K/work | $100K/work | $125K/work | $150K/work |
| 10,000 | $500M | $750M | $1B | $1.25B | $1.5B |
| 100,000 | $5B | $7.5B | $10B | $12.5B | $15B |
| 250,000 | $12.5B | $18.75B | $25B | $31.25B | $37.5B |
| 500,000 | $25B | $37.5B | $50B | $62.5B | $75B |
| 750,000 | $37.5B | $56.25B | $75B | $93.75B | $112.5B |
| 1,000,000 | $50B | $75B | $100B | $125B | $150B |
| 1,500,000 | $75B | $112.5B | $150B | $187.5B | $225B |
| 2,000,000 | $100B | $150B | $200B | $250B | $300B |
| 3,000,000 | $150B | $225B | $300B | $375B | $450B |
| 4,000,000 | $200B | $300B | $400B | $500B | $600B |
| 5,000,000 | $250B | $375B | $500B | $625B | $750B |
| 6,000,000 | $300B | $450B | $600B | $750B | $900B |
For all of the book authors’ lawsuits, we should expect the plaintiffs to use the one-two punch of (1) the Shadow Library Strategy and (2) the Class Action Strategy That Includes Statutory Damages.
If you look at the map of U.S. copyright lawsuits, many involve book authors:
3. anti-rag strategy
The next strategy Susman Godrey is advancing is the Anti-RAG Strategy against so-called “answer engines.” The recently filed lawsuit by Encyclopaedia Britannica v. Perplexity is the best example. Susman had already been advancing this Anti-RAG Strategy against Microsoft in the New York Times lawsuit. No court has yet considered an infringement claim or the fair use and other defenses in the context of RAG. It’s somewhat reminiscent of the early lawsuit against Google News that eventually settled. But, for RAG, some answer engines might be simply reporting facts without copying expression. (I wrote some preliminary thoughts on this controversial issue here.)
NYT v. Microsoft allegation:
4. Antitrust Strategy
Susman Godfrey just opened up its playbook and is now representing Penske Media in its antitrust lawsuit against Google for its AI mode and conditions for being included in Google search. (Susman also represents Chegg in a similar antitrust suit against Google.)
Interestingly, the only lawsuit in which Susman Godfrey has advanced a DMCA CMI removal claim is the New York Times lawsuit. Unlike the Kadrey book authors, the Bartz book authors did not advance this claim, and none of the other book author suits filed by Susman Godrey includes it.
Why the susman godfrey playbook is a brilliant strategy
While the issue of fair use in AI training gets the most public and scholarly attention, Susman Godfrey has developed a playbook, especially Strategies 1 and 2, that does not depend on prevailing on the fair use issue.
In fact, in Bartz, it lost on fair use yet still secured the largest proposed settlement in U.S. copyright history.
If Susman Godfrey replicates this strategy in one or more of the lawsuits it is involved in against OpenAI, Microsoft, and Perplexity AI—especially the former two companies in the MDL Litigation where Susman is interim lead class counsel–the strategy will be a monumental success.
And, if it prevails in the antitrust lawsuit against Google, it will be icing on the AI cake.
Video of Justin Nelson after Dominion settlement
Justin Nelson is the lead interim class counsel in MDL Litigation v. OpenAI, Microsoft.
Related Stories