Unclean Hands

Unclean Hands

Unclean Hands is a principle from Anglo-American law: someone who has behaved unfairly themselves gets no support from the court against their opponent. In AI disputes, the objection comes up when, for example, a plaintiff obtained their own data in a questionable manner.

Unclean Hands is a rule from the legal systems of the USA and the United Kingdom. Translated, the term means “unreine Hände” (unclean hands). What it means is: someone who has themselves acted unfairly cannot demand special help from the court against someone else. The underlying idea is old and simple. Whoever demands justice must have stayed clean themselves in that same matter. A thief who gets robbed can certainly file a police report – but he cannot demand that a court help him divide up the loot.

An objection that can topple entire lawsuits

Unclean Hands is not an accusation brought by the prosecution. It is a defense. The defendant is essentially saying: take a look first at what the plaintiff himself has done. If the court goes along with this, the plaintiff loses – without the actual matter in dispute ever being decided.

That is precisely what makes the objection so effective. It forces both sides to disclose their own conduct. In proceedings involving software, data, and AI, this is a sharp sword. Many companies collected data in their early stages whose origins they would rather not discuss today.

The boundary is important: the misconduct must be connected to the subject matter of the dispute. A company that had trouble with a tax matter years ago does not automatically lose a copyright lawsuit because of it. The court examines whether the accusation is substantively related to the case.

The difference between damages and injunctive relief

To understand Unclean Hands, one must distinguish between two types of claims. One is money: damages for a harm suffered. The other is an order to do or refrain from doing something. A court can, for instance, prohibit the continued offering of a particular model. Such orders are called “equitable remedies” in English law, that is, remedies based on fairness.

Unclean Hands traditionally applies only to this second group. The background is historical. In England, there used to be separate courts of equity alongside the ordinary courts. They were meant to step in where strict rules led to unfair outcomes. Whoever wanted this additional fairness had to have been fair themselves.

In practice, it works like this: the defendant presents the accusations and must substantiate them. The court weighs how serious the misconduct is and how closely it relates to the dispute. There is no fixed checklist for this. Judges have considerable discretion, which makes it hard for litigants to predict outcomes.

Where the term comes up in AI lawsuits

Since 2023, numerous lawsuits against providers of AI models have been proceeding in the USA. Publishers, authors, and image agencies accuse them of using works for training without permission. Training means: a model analyzes huge amounts of text in order to learn patterns in language.

In these proceedings, AI companies occasionally argue Unclean Hands. One accusation, for example, is that a plaintiff automatically scraped content from other websites themselves in order to build up their own database. Another concerns terms of use: whoever tested a product in violation of the agreed rules in order to gather evidence is not entirely clean themselves.

For investors, this is relevant because such proceedings can determine outcomes worth billions. A successful Unclean Hands objection can avert an impending injunction. A common misconception, however, is that the objection renders the lawsuit completely worthless. The plaintiff can still receive damages – just not the injunction that would truly hurt the defendant. In Germany, the doctrine does not exist under this name. Similar ideas here are embedded in the principle of good faith (Treu und Glauben).

Subscribe free. Unsubscribe the second it sucks.

High-signal news across AI, business, UX, and tech. Every morning.