Robots and IP: Protecting Faces, Expressions and Vocalizations

“The case illustrates the relatively high degree of overlap that must exist between the protectable elements of different works for a copyright holder to prevail under the substantial similarity test.”

Preventing others from copying your robot’s AI-driven face, expressions and vocalizations requires a comprehensive intellectual property strategy. That’s one of the takeaways from a pending dispute between robot makers as described in Digital Dream Labs, LLC v. Living Technology (Shenzhen) Co. (pending in the Western District of Pennsylvania). The case involves plaintiff DDL, which owns registered copyrights in desktop humanoid-vehicle hybrid robots called COZMO and VECTOR (see below left and middle), and defendant Living.AI, whose headphone-wearing, skateboard riding, humanoid robot called EMO (below right) is alleged by DDL to infringe its copyrights. Both companies reportedly deployed AI software on their robots that selects graphical animations and sounds to output based on the robot’s reactions with its environment and user.

COZMO, VECTOR, and EMO Robots (not to scale) Source:

DDL sued Living.AI for copyright infringement (among other claims). Living.AI moved to dismiss. Upon considering the parties’ submission and a live demonstration of the robots, the court denied Living.AI’s motion with respect to DDL’s audiovisual copyright infringement claim after finding that a rational lay person would view the audiovisual components of the parties’ competing robot faces as being “substantially similar” (a necessary element of copyright infringement discussed below), but granted Living.AI’s motion to dismiss with respect to DDL’s sculptural copyright infringement claim after finding that a lay person would not view EMO’s physical structural features as being substantially similar to the COZMO and VECTOR copyrighted features. Digital Dream Labs, LLC v. Living Technology (Shenzhen) Co., slip op., No. 2:2020-cv-01500 (W.D. Pa. Feb. 28, 2022).

Copyright Infringement

Under U.S. copyright laws, a graphical, synthetic, anthropomorphized “face” displayed on a screen and its related “vocal” sounds may be protected as an audiovisual work, which is defined as a work consisting of a series of related images that are intrinsically intended to be shown by the use of machines, or devices such as projectors, viewers, or electronic equipment, together with accompanying sounds, if any, regardless of the nature of the material objects…in which the works are embodied (17 USC 101). That said, not every humanoid robot capable of expressing human-like emotions and interacting with the world is a protectable work under copyright laws. Only those that involve some element of original creative expression (not to be confused with facial expressions discussed here) recognized under the law may be protected. The challenge for creators (and their lawyers) is to identify those protectable elements in a robot (if any), seek copyright protection for them, and then routinely checking for unlawful uses of the elements by others.

To establish a claim of copyright infringement of an audiovisual work based on copying, a plaintiff must show: (1) ownership of a valid copyright; and (2) unauthorized copying of original elements of the registered work. The “unauthorized copying” element of infringement requires a showing of (i) actual copying and (ii) material appropriation of the copyrighted work. Both parts (i) and (ii) must be shown because not all instances of actual copying give rise to liability, and, conversely, without proof of actual copying, the degree of similarity between two audiovisual works is immaterial. A “material appropriation” occurs when there is “substantial similarity” between the work alleged to be infringing and the protectable elements of the copyrighted work.

In determining substantial similarity, a court looks at how the copyrighted and the allegedly infringing works would appear to an ordinary (lay) observer viewing them side by side, and considers whether the observer would believe that the copying was of protectable aspects of the copyrighted work. The court must “maintain focus on the protected elements of the copyrighted work, not the prominence of any such elements in the defendant’s work.” See Digital Dream Labs, LLC v. Living Technology (Shenzhen) Co., slip op., No. 2:2020-cv-01500 (W.D. Pa. Feb. 28, 2022) (citations omitted). Courts must take into account a work’s “total concept and feel.” But, where, as here, there is a mix of protected and unprotected elements, the first step is to identify and exclude (filter) from the substantial similarity analysis any unprotected material. (Note: the substantial similarity test is expressed differently in other jurisdictions).

Audiovisual Face/Voice Functions of a Robot Are ‘Protectable Elements’

Generally, the idea of robots displaying humanoid emotions on their “faces” with accompanying sounds is not protectable under copyright; nor is using only eyes to convey emotions by changing their size and direction, as the robots at issue here did. That is because, as the court noted, most people would recognize the emotion of sadness when eyes are tilted downwards, the expression of anger or frustration when eyes are pushed together as a result of a furrowed brow, the expression of surprise when eyes open wide and move upward, or the expression of sleepiness when eyes slowly close. Under the “scènes à faire” doctrine, the idea of a robot responding to questions or providing information, such as the weather, is unprotectable as “stock and standard” stuff. “Such abilities are expected features that a humanoid robot may be expected to have,” the court here concluded.

Here, however, the court pointed out that the expression of emotions through VECTOR’s and COZMO’s “particularized eyes” (shape, size, color, location, movement), along with the accompanying particularized sound, created a protectable expression of the robots’ emotions. Specifically, the court noted that when VECTOR indicates that it is raining, raindrops begin to fall on its eyes and the drops bounce off the top of its eyes with a splash and trickle down its face, VECTOR then partially closes and lowers its eyes as the rain continues. When VECTOR takes pictures, its eyes turn from squares to circles and rotate like a focusing camera lens, and then blink when taking a picture. In both instances, the court noted, DDL could have chosen other animations, suggesting that the ones it chose were not stock or standard and thus did not “monopolize an underlying idea.”

Turning to the “substantial similarity” question, the court concluded that a reasonable jury could find a substantial similarity between the VECTOR and COZMO robots on the one hand, and the EMO robot on the other. The court observed that each of VECTOR, COZMO, and EMO’s faces displayed specific animations and graphics in response to different external stimuli on a screen that is located on the front plane of their heads. The display of an animation or graphics is typically accompanied by a related sound, such as, for example, when indicating that it is raining, a rainfall noise is emitted. All three robots have glowing, blue/green, rectangular (at times square), eyes with rounded edges that fade towards the edge of the eyes. Their eyes appear approximately on the center of the face, with a black background. When expressing an emotion, each robot outputs a metallic, high-pitched, and over-exaggerated sound to accompany the expression of each emotion. When conveying information, such as weather or time, the robots have similar aesthetic animations. Based on “the total concept and feel” of their screens, the court found that an ordinary observer may perceive that EMO’s face copied protected elements of VECTOR’s and COZMO’s face.

Sculptural Features of a Robot Are Protectable Elements

The court noted that the protectable elements of DDL’s desktop robots are their combination and arrangement of appendages, accessories, camera location, and shape, as well as overall size. Shape, according to the court, was the most distinguishing feature between the robots, and pointed out that EMO is about twice as tall as VECTOR and COZMO. But also, their appendages “differ significantly”: VECTOR and COZMO each have two sets of tractor treads, whereas EMO has two leg-like appendages with cube-like feet; EMO has no arm-like appendages, whereas both VECTOR and COZMO have forklift-like arms that join together in the front and extend from the front of their tractor treads to behind their heads). EMO, the court observed, has a cuboid head balanced on two thin legs, whereas VECTOR and COZMO have a wide base and forklift-like arms that surround the cuboid head. The VECTOR and COZMO heads are significantly smaller than EMO’s head. Moreover, EMO wears over-the-ear headphones, whereas VECTOR and COZMO do not. The ratio of the screen-to-head differs among the robots. EMO’s screen is centered in the middle of the head, contains an outer ring that parallels the outer shape of head, and has a camera indicated by a small half-circle at the top center of its forehead. In contrast, both VECTOR and COZMO have screens that appear to extend to the edge of the top and bottom of the head and have a camera at the bottom part of the face inside a long rectangular bar, resembling an open mouth.

Based on these differences, the court found that a lay person would not believe that the EMO robot copied protectable aspects of DDL’s copyrighted three-dimensional sculptural works reflected in VECTOR and COZMO. As a result, the court found that no reasonable jury could find that the three-dimensional sculptures are substantially similar between the parties’ robots.

The Takeaway

This case illustrates the benefits of protecting various hardware, software, and aesthetic aspects of one’s physical robot by using available IP laws, such as copyrights (but also, where applicable, patents, trademarks (including trade dress), and (where feasible) trade secrets; the DDL case, for example, also raised trademark and trade dress claims in addition to copyright claims). This case also illustrates the relatively high degree of overlap that must exist between the protectable elements of different works for a copyright holder to prevail under the substantial similarity test. Looking beyond this case, as AI-generated, synthetic (fake) faces find their way into more use cases beyond toy robots, it will be interesting to see how companies employ IP laws to successfully protect their creations and how courts will decide what are the protectable features of artificial personas.


Warning & Disclaimer: The pages, articles and comments on do not constitute legal advice, nor do they create any attorney-client relationship. The articles published express the personal opinion and views of the author as of the time of publication and should not be attributed to the author’s employer, clients or the sponsors of Read more.

Join the Discussion

One comment so far. Add my comment.

  • [Avatar for B]
    April 13, 2022 10:49 am

    This is a well-written and fun article, Mssr. Higgins

Add Comment

Your email address will not be published.