Elizabeth A. Corradino is quoted in a WTR article titled, "Netflix’s new genAI guidance: a “sensible” step towards respecting IP and reducing risk?" 

In this week’s opinion column, IP and technology experts offer their thoughts on Netflix’s new guidance for film production partners using generative AI tools. 

Generative AI’s capacity to create impressive, realistic-looking footage, that might otherwise have been prohibitively expensive or technologically unfeasible to shoot, has proved enticing, but controversial, in the filmmaking world.

In July, for example, Netflix’s co-CEO Ted Sarandos revealed that the Argentinian crew behind its hit sci-fi series, The Eternaut, had used the technology to create a sequence in which a building appears to collapse in Buenos Aires. “Using AI-powered tools, they were able to achieve an amazing result with remarkable speed and, in fact, that VFX sequence was completed ten times faster than it could have been completed with traditional VFX tools and work flows,” Sarandos told analysts in an earnings call. 

“The creators were thrilled with the result. We were thrilled with the result, and more importantly the audience was thrilled with the result,” he enthused, arguing that the creative opportunities opened up by these tools are “endlessly exciting”.

However, for others in the industry, the possibility that filmmakers might start relying more and more on this transformative technology has provoked more fear than excitement. 

Concerns that generative AI may be used to replace actors or create digital replicas of them after death were among those that inspired the 2023 Hollywood strikes, for example. The use of copyrighted works to develop these tools was also a concern. 

“We think it is absolutely vital that this negotiation protects not just our likenesses, but makes sure we are well compensated when any of our work is used to train AI,” urged a letter to SAG-AFTRA leadership, during its contract negotiations with the Alliance of Motion Picture and Television Producers. 

More recently, accusations that those involved in the production of Netflix’s true crime documentary What Jennifer Did may have used generative AI to create or manipulate images, presented as real archive photos, caused a stir. “This isn't inventing a fictional narrative for the sake of entertainment — this is tinkering with the fabric of reality itself to manipulate a true story that actually happened,” Futurism’s Victor Tangermann argued at the time. 

Executive producer Jeremy Grimaldi denied that the photos in the documentary were manipulated by AI. “Any filmmaker will use different tools, like Photoshop, in films,” he told the Toronto Star. “The photos of Jennifer are real photos of her. The foreground is exactly her. The background has [just] been anonymized to protect the source,” he claimed.  

The use of generative AI in filmmaking clearly raises a spectrum of ethical questions, including those around intellectual property and rights of publicity. It also poses a significant legal and reputational risk for Netflix, as it continues to add to its online streaming catalogue. 

Netflix too appears to have recognised this. 

Last month, the streaming service released guidance for ‘Using Generative AI in Content Production’ as part of its Partner Help Center.

The new guidance is intended to help filmmakers, production partners and vendors “understand when and how to use GenAI tools in production”, it explains. “It also offers a practical tool for assessing and enabling confident GenAI use when producing content for Netflix.” 

“To support global productions and stay aligned with best practices, we expect all production partners to share any intended use of GenAI with their Netflix contact,” the guidance states.

“Most low-risk use cases that follow the guiding principles below are unlikely to require legal review,” it clarifies. “However, if the output includes final deliverables, talent likeness, personal data, or third-party IP, written approval will be required before you proceed.” 

Among other things, the guidance urges production partners:

• not to train or fine-tune models using content that has not been properly cleared;
• not to create digital replicas of performers or their voices without “explicit and documented consent” and in compliance with guild requirements (where applicable);
• not to use generative AI to create “main characters, key visual elements, or fictional settings that are central to the story” without written approval;
• to avoid creating content that “could be mistaken for real events, people or statements if they never actually occurred”; and
• to avoid using inputs “that reference copyrighted materials or likenesses of public figures or deceased individuals without appropriate permissions”.

“Overall, these guidelines are a sensible component of Netflix’s strategy for managing risk,” remarks Mark Lezama, co-chair of Knobbe Martens’ copyright litigation practice. This is true both for managing legal risk and for managing the risk of negative consumer reaction to using generative AI outputs, he adds. 

However, whether the guidelines are effective will “hinge on the compliance of Netflix’s partners”, he argues. Indeed, the guidance makes no mention of how Netflix intends to enforce these standards, if at all. “Any contractual or other remedies Netflix may have against a partner that does not abide by these disclosure guidelines might not be enough to mitigate the cost of a lawsuit,” Lezama worries. 

“In practice, it may be difficult for a filmmaker to know whether the AI tools that they are using adhere to Netflix’s criteria,” he continues. “The guidelines tell partners to consult their Netflix contact, but it is unclear what the practical solution will be,” he stresses. 

“Nonetheless, the guidelines do promote early disclosure and dialogue, which always helps when managing risk,” he admits. 

Elizabeth Corradino, chair of Moses Singer’s IP, entertainment/media and technology practice group, offers a more favourable view of the guidance. 

“By requiring written approval before using generative AI to generate key creative elements, the guidelines preserve the unique voice of creators and help prevent over-reliance on automated tools for story-critical elements,” she argues.

“The explicit prohibition against referencing copyrighted materials or the likeness of public figures and deceased individuals is a critical safeguard that demonstrates respect for intellectual property and name, image and likeness rights,” she notes, approvingly. 

More practically, “the inclusion of clear examples makes it easier for creative teams to responsibly navigate the evolving landscape of generative AI”. 

While generative AI tools may be relatively novel, policies intended to address the ethical and IP implications of emerging technologies are not, Corradino highlights.

She notes, for example, that:

• the adoption of computer-generated imagery (CGI) led to new industry standards and IP considerations regarding digital assets;
• the transition from analog to digital editing mandated the creation of new workflows, as well as retraining and updating provisions in agreements with editors and post-production personnel; and
• the use of actors’ performances to animate digital characters (eg, Andy Serkis as Gollum in Lord of the Rings) raised questions about credit, compensation and rights over digital likenesses, all of which, she notes, “required answers and adjustments”. 

“Though generative AI itself may be new, content production policies are a well-established part of the entertainment industry,” Corradino underlines. 

Netflix’s guidance may concern novel technology, but policies like it are a common feature of the industry, she says, “guiding the creation of content in accordance with the legal, ethical and business considerations of the time”. 

___________________________________________

**** This article was originally published in World Trademark Review and is republished with permission.