It was a busy week in AI copyright litigation in the United States with three new decisions released in litigation involving Anthropic, OpenAI, and Microsoft. One involved a failed attempt by music publishers to obtain an injunction against Anthropic from using lyrics to train its AI system, Claude. Two of the cases highlight pleading requirements in which certain claims of infringement were dismissed against OpenAI, Microsoft, and Anthropic.
Concord Music v Anthropic
The Concord Music v Anthropic action arises out of the Anthropic’s alleged use of Plaintiffs’ copyrighted song lyrics to train a generative artificial intelligence (“AI”) model. Plaintiffs, eight music publishing companies who own or control exclusive rights to millions of musical compositions, sought a preliminary injunction that prohibited Anthropic from using the copyrighted lyrics for training purposes. The motion was denied, essentially for two reasons.
First, the relief claimed was extremely broad covering undefined present and future songs. “The enormous and seemingly ever-expanding scope of Works included in the requested injunction raises significant concerns regarding enforceability and manageability.” “The Court does not have unfettered discretion to issue an injunction that may be overbroad because it includes works for which Publishers may not have established ownership or control.”
Second, the publishers were unable to demonstrate irreparable harm arising from using the publishers’ copyrighted lyrics for AI training purposes and the publishers did not pursue irreparable harm based on outputs generated using Claude.
An argument advanced was the publishers’ loss of control over the uses of their works resulted in irreparable harm. The court was not satisfied that the evidence adduced on the motion established such harm.
Publishers contend that Anthropic’s ongoing use of the copyrighted works will cause them to suffer reputational harm by depriving them of control over the works, denying them credit and goodwill associated with the works, and harming their reputations through the creation of unauthorized derivative works. A loss of control over business reputation or damage to customer goodwill “could constitute irreparable harm” where supported by evidence. ..For the reasons discussed below, Publishers have not made this showing here.
The Court notes that the reputational harm described by Publishers appears largely related to Claude’s outputs rather than the use of the Works for training purposes alone.
For example, Publishers claim that “Claude routinely reproduces, distributes, and displays Publishers’ copyrighted lyrics in its output but fails to identify Publishers or their songwriters as the source.” And “Claude frequently generates output that combines portions of the Works with other lyrics or text, often in ways inconsistent with and inimical to authorial intent – and for which the songwriter would never have granted a license.” The parties’ stipulation addressed and resolved Publishers’ motion for injunctive relief to the extent it concerns Claude’s output.
Publishers have not demonstrated reputational harm based on the use of the Works as training input. Publishers cite to several declarations, which are largely duplicative of each other.
The declarations discuss the effect of unlicensed use of copyrighted works as “damaging” or “harmful” generally, but they fail to identify any specific harm…. Although the Court acknowledges the value an artist places on control of their own works, none of the declarations explain how or why the use of the Works to train Claude has been, or will be, “enormously damaging” to Publishers or songwriters. To a certain extent, all copyright infringement involves a loss of control. But accepting Publishers’ argument on this record would imply automatic entitlement to injunctive relief in all copyright cases, a premise the Ninth Circuit has rejected… A preliminary injunction must be “grounded in evidence,” not based on what harm Publishers “might suffer.”
The publishers also contended that Anthropic’s use of copyrighted lyrics would erode the value of the works by undermining the licensing market, damaging the publishers’ position to negotiate future training licenses with AI developers, and harming the publishers’ relationships with songwriters. In effect, the publishers argued that output containing their works and the use of the works to train Claude affected the publishers’ position both (1) in the existing licensing market and (2) in the emerging market for licensing copyrighted works for use by AI developers.
The court also rejected the evidence on this argument. It held that the “Publishers have not submitted any evidence that Anthropic’s use of lyrics to train Claude reduces license fees with lyric aggregators, lyric websites, or other existing licensees, which provide entirely different services and do not compete with Claude”.
[Publishers] do not demonstrate how using the Works to train Claude is affecting – let alone diminishing – the value of any of the Works. Nor do they show that training Claude with the Works is harming Publishers’ negotiating position vis-à-vis new or existing licensees like lyrics aggregators. For example, Publishers have not identified any lost licensing deal or any licensing arrangement that Publishers have had to renegotiate on less favorable terms…
The publishers also argued that Anthropic’s use of the works would negatively impact the emerging market for AI training licenses and diminish their ability to negotiate such licenses in the future. They argued that “[i]f Anthropic “continues to exploit and devalue the Works, even if only during the pendency of this case, its conduct will become entrenched in the AI industry and the public consciousness.” They further contended that Anthropic’s use “will permanently undermine Publishers’ leverage in future negotiations with AI developers to license lyrics as training data.”
The court held that the evidence cited by the publishers was insufficient to support this theory. According to the court:
They fail to provide details or specifics regarding how, if at all, Anthropic’s use of the Works to train Claude has affected their respective abilities to negotiate training licenses with other AI developers, or how it will inflict harm on the emerging licensing market. Publishers have not demonstrated a likelihood of harm arising from Anthropic’s use of the Works for training purposes.
The publishers also contended that “courts must treat the use of copyrighted works in emerging technology markets with caution and care”.
The court did not disagree with this, but considered this favored Anthropic as it would have required the court to consider the fair use defense and how that could affect the licensing market for works. The court was not prepared to do so on the preliminary injunction motion.
…Publishers fail to explain how this general proposition entitles them to relief. To the contrary, these cases demonstrate that emerging technologies often test the bounds and principles of copyright law. Here, it is an open question whether training generative AI models with copyrighted material is infringement or fair use. By seeking a preliminary injunction, Publishers are essentially asking the Court to define the contours of a licensing market for AI training where the threshold question of fair use remains unsettled. The Court declines to award Publishers the extraordinary relief of a preliminary injunction based on legal rights (here, licensing rights) that have not yet been established.
Procedural motions to strike
In a separate decision, the district court in the Concord Music v Antrhopic case, struck the publishers’ claims for contributory infringement and vicarious liability claims related to outputs generated by Claude. In essence, these claims were dismissed, with leave to amend, as the publishers’ claim failed to plead evidence of direct infringement by users of the AI model or that Anthropic had sufficient knowledge of such infringements to become liable. Claims based on infringement under the DMCA by the removal of copyright management information (CMI) were also dismissed, with leave to amend.
A separate district court decision also dismissed several claims against OpenAI and Microsoft in three suits brought against them. For reasons that the court will release later, the court dismissed various claims including a claim based on unfair competition (presumably based on copyright preemption) and certain DMCA claims related to CMI.
1 comment