The practices associated with intellectual property indemnity can be traced in part to Article 2 of the Uniform Commercial Code. At the dawn of the computer age, practitioners searched for legal models that they could use for transactions in intangible rights and products such as computer software. Although computer software did not fit easily into the “sale of goods” paradigm, analogies to the familiar rules governing sales of goods were inevitable.
Lurking in the lower reaches of Article 2 of the UCC, one finds an implied warranty of non-infringement in Section 2-312(2):
Unless otherwise agreed, a seller that is a merchant regularly dealing in goods of the kind warrants that the goods shall be delivered free of the rightful claim of any third person by way of infringement or the like but a buyer that furnishes specifications to the seller must hold the seller harmless against any such claim that arises out of compliance with the specifications.
In the context of a sale of goods at the time Article 2 was drafted, a non-infringement warranty made good economic sense. The only form of intellectual property likely to be of concern to the purchaser of goods was patent protection. In the mid-twentieth century, when Article 2 of the UCC was adopted, patents were the disfavored stepchildren of the federal courts. A high percentage of patents were held to be invalid, and the damages accorded to those found to be valid were often limited to a modest royalty. Furthermore, patent rights are—as a rule—exhausted upon the first sale of a product and far fewer patents were being issued. Additionally, products were less complex, typically falling into only a single engineering domain, with correspondingly fewer points of intersection with issued patents. For all these reasons, the risk that the ultimate purchaser of a product would be sued for patent infringement was very remote, and even if the purchaser were sued, the damages would be only a tiny fraction of the purchase price—the average and median for all products was between five and seven percent.
In this context, a product manufacturer could provide an implied warranty of non-infringement with very little risk beyond whatever modest risk of infringement the manufacturer had already incurred by manufacturing the product itself. Product purchasers did not typically insist on an indemnity beyond the implied warranty , perhaps because the risk was perceived as too remote to be worth worrying about. The exhaustion doctrine would—in many cases—cause the claim to be made against the manufacturer rather than the user. Therefore, they would assume—probably quite rightly—that the manufacturer would “stand behind its product” and obtain the required license if an infringement claim were actually made.
With this background, and without giving the matter too much additional thought, it did not seem unreasonable for the lawyers representing software and other technology providers to provide a warranty of non-infringement governing their clients’ products as well. Such a warranty was certainly consistent with established practice as embodied in UCC Article 2. It therefore would have met the expectations of the marketplace. Furthermore, the risk of infringement liability seemed manageable. At that time, virtually no one believed that computer software would be patentable, and absolutely no one imagined that business methods would be patentable. The risks of trade secret or copyright infringement liability are more uniquely within the control of the software provider since each of them requires an element of intent or a near equivalent. By adopting appropriate internal controls and standards, sometimes even including a “clean room,” a software provider ought to be able to minimize its infringement risk.
As licensors, software providers had an additional consideration of the desire to control litigation concerning rights in the product. At that time, protection for computer software was in a state of flux. Copyright in computer software expanded in the years following the Copyright Act of 1976, but doubt persisted as to the extent of its protection. For instance, in following decades, the courts were faced difficult questions regarding whether copyright protected operating systems that communicate only with machines and whether it protected the “look and feel” of the user interface generated by the software. These were considered weighty issues going to the heart of the value represented by computer software. Accordingly, the software providers did not want to risk having them decided in litigation against their licensees, whose interest in broad protection would likely be less than the interest of the software providers.
In response to this concern, software providers migrated the infringement issue from the warranty clauses of software license agreements to indemnity provisions. As quid pro quo for indemnity, the licensor obtained prompt notice of the claim and full authority to defend or settle the case on behalf of the licensee. To mitigate their risk, most software licensors further provided that their obligation—and the licensee’s sole remedy—was to (a) obtain for the licensee the right to use the licensed product; (b) modify the product to make it non-infringing; or—failing (a) or (b)—(c) to terminate the license and refund a pro-rated portion of the purchase price depending on how much of the license term had been exhausted at the date of termination.
Clauses along these lines became a de facto standard in the software industry and persisted for many years. They were hardly perfect from the standpoint of licensees. Option (c) could be very detrimental in the case of software that was critical to a business, Even if comparable non-infringing software were available, the end user would be in the position of having to find, implement, and migrate all data or customizations to the new system. This could be expensive, risky and time-consuming. If no comparable non-infringing software were available and the patentee refused to offer a license on commercially reasonably terms, a business that relied on the software could be crippled. Even option (b) could be problematic if the required modifications degraded functionality or compatibility. As a result, licensees with sufficient bargaining power would nibble at the edges of such indemnity clauses, sometimes allowing full refunds and often providing that (b) could only be exercised if functionality were not compromised. Nonetheless, the basic pattern—limited indemnity in exchange for control over litigation—became well-fixed in many practitioners’ minds.
As so often happens in the law, the basic practices that emerged in one field—the software industry—migrated to other industries as well. Intellectual property indemnification clauses were inserted into virtually any agreement in which the parties foresaw a risk of intellectual property infringement.