Deepfake images can be easily manipulated to remove ‘AI fingerprints’ making it difficult to conclude if an image has been AI generated, a study has found.

It is also possible for hackers to change an image’s AI fingerprint so it falsely appears to come from a different model. This could be used to wrongly blame legitimate tech companies for harmful images their systems never actually created.

Experts say improvements in AI fingerprinting techniques combined with watermarking of AI-generated images would strengthen the detection of deepfakes.

Generative AI is now capable of creating images nearly indistinguishable from real photos, raising concerns about the use of these technologies for scams and misinformation campaigns.

Hinder Investigations
One promising approach to mitigate these risks is AI fingerprinting—a group of techniques that detect unique, invisible traces that AI models leave in their images, which helps identify the specific generator that produced them.

Removing fingerprints would hinder forensic investigations into deepfakes. Scientists from the University of Edinburgh have found that these fingerprints can be removed or manipulated using various modes of attack.

In the first part of the study, a security evaluation of fingerprinting techniques for generative AI was performed. Adversarial attacks were then developed that aimed to remove or forge fingerprints across a range of threat scenarios.

These scenarios ranged from powerful attackers with full access to the inner workings of the image generator to low-resource attackers with no special access.

The scientists simulated these attacks on 12 image generators and 14 fingerprinting methods in the largest evaluation of such techniques to date.

Many fingerprinting methods were found to achieve high accuracy in detecting unaltered deepfake images, but performance drops dramatically once the image is attacked.

'Smudging Fingerprints'
Fingerprint removal was found to be highly effective, often achieving more than 80% success for attackers with full knowledge of an image generator and just over 50% for simple attacks with no knowledge of the generator’s inner workings.

In several cases, simple changes to an image such as JPEG compression, resizing or blurring are enough to 'smudge' the fingerprints.

Fingerprint forgery—misrepresentation of the AI model used to generate the image—was less effective than removal overall, but half of the image generators evaluated were vulnerable to this kind of attack.

All attacks were imperceptible to the human eye, leaving no visible evidence on the images. None of the evaluated fingerprinting techniques delivered both high accuracy and resistance to attack across all threat scenarios.

By pinpointing where and why current approaches fail, it should be possible to build stronger methods of deepfake detection, especially when paired with watermarking – the process of embedding a hidden digital signature into AI-generated content.

The findings from this work were peer reviewed and will be presented at the IEEE Conference on Secure and Trustworthy Machine Learning (SaTML) in Munich this week. A copy of the final version of the paper is available here.

This work was supported by the Edinburgh International Data Facility, the Data-Driven Innovation Programme and the Generative AI Laboratory, all at the University of Edinburgh.

Kai Yao, PhD student at the University of Edinburgh’s School of Informatics and author on the paper, said: “We were surprised to find just how fragile these AI fingerprints truly are. We expected that sophisticated attacks would be effective, but seeing that simple, everyday image edits could effectively ‘smudge’ the forensic evidence was a real wake-up call. It suggests that many of the deepfake detection methods based on image fingerprinting might fail the moment an image is shared or edited in the real world.”

Dr Marc Juarez, Lecturer in Cyber Security and Privacy at the University of Edinburgh and also an author on the paper, said: "Deploying these techniques without considering the threats they face could give a false sense of security. If fingerprinting is to be used to hold bad actors accountable, it must ensure that fingerprints cannot be easily removed or forged, as any accountability tool will itself become a target for attack. The community must therefore move beyond optimising for performance alone and incorporate adversarial robustness into their evaluation methodology."

With the Federal Communications Commission working on new rules for the deployment of NextGen TV, next year promises to be an important one for both the future of ATSC 3.0 in general and for the security protections that are a key component of the new broadcast standard.

“For us, content protection is table stakes in terms of broadcasters being able to compete in digital environments,” explained Ann Schelle, managing director of Pearl TV, a consortium of major station groups backing the NextGen TV rollout. “To compete with streaming and other digital media, we need the same tool kit, the same content protection, they have.”

While ATSC 1.0 lacked encryption and robust content security, the content-security features embedded in ATSC 3.0 are particularly important, as NextGen TV is being deployed when such protections are ubiquitous in the digital and streaming world, added Robert (Rob) Folliard, Gray Media’s senior vice president, government relations and distribution.

“DRM [digital rights management] is critical for high-value content from the NFL, MLB, the NBA and all the other leagues, because they lose billions of dollars each year from piracy,” he explained. “That is a problem for us—live sports drives the broadcast model, and we are the only video provider who doesn’t have it. The NFL doesn’t have to worry about it when they do a deal with Amazon or YouTube. If we don’t have content protection, that is going to be a strike against us in the next round of negotiations.”

New Rules, Old Issues
Many of the ongoing arguments and controversies over content security in the digital world can be found in the FCC’s electronic filing system, where more than 7,000 comments have been tallied in response to the agency’s plans to develop new NextGen TV rules.

While Pearl TV and larger broadcasters have supported the enhanced security and content protection offered by the 3.0 standard, some LPTV stations and smaller station owners have raised concerns about the high costs of implementing security solutions.

These filings have been particularly critical of the ATSC 3.0 Security Authority (A3SA), which was created by broadcast TV networks and station groups to manage the security framework for NextGen TV, including digital rights management and content-protection protocols.

The LPTV Broadcasters Association, for example, has argued that the expense of transitioning to 3.0 broadcasts would force stations out of business and that the costs of obtaining “signal-signing” certificates to verify the 3.0 broadcast signals are “excessive.”

Separately, thousands of filings from consumers and advocacy groups like Public Knowledge have also complained that encryption and signal-signing certificates undermine the public’s ability to access free broadcast signals.

“Introducing encryption and Digital Rights Management (DRM) undermines this foundation by transforming open access into conditional, license-controlled access,” Public Knowledge told the FCC. “DRM and the associated A3SA certification system would allow private entities to determine which manufacturers—and therefore which consumers—may access free over-the-air content, effectively privatizing what has always been a public medium,” as well as increasing the cost of devices, the group argued.

Schelle adamantly disputes those reactions, arguing that much of the opposition is either based on a fundamental misunderstanding of 3.0 security or that complaints come from groups simply opposed to ATSC 3.0.

The LPTV Broadcasters Association and “Weigel [Broadcasting] are not for NextGen TV, full stop,” she said. “‘SuperFrank’ [Copsidas, the founder of the LPTV Broadcasters Association] is in favor of 5G broadcasting. Weigel would prefer that we stay on 1.0…so, of course, they are going to…glom onto these issues and blow them up.”

Schelle and Joe St. Jean, managing director of A3SA, also stressed that opponents have misinterpreted his group’s workings and the role of security in the rollout of NextGen TV.

“When you hear about content protection, people often think that they are losing rights,” said St. Jean. “That is not the case here. We are not trying to box out consumers from viewing the content they could otherwise access…We want consumers to use DVR and home-networking functionality…This is a free environment and will continue to be free…Broadcasters are in the eyeball business, and we want consumers to view the content.”

More specifically, he said, it’s important to separate two security-related issues: the DRM solutions used to protect content and signal-signing certificates to authenticate broadcast signals.

“Content protection is described in the ATSC 3.0 spec, but it doesn’t describe implementation, which is what A3SA did,” St. Jean explained. “They created an implementation, and its use is optional. We are not looking to enforce content protection on folks or make money off it.”

For the A3SA implementation, they turned to Google’s Widevine DRM solution, which is used in tens of billions of devices for content protection, Schelle added. “It is enabled in all the TV manufacturers, it works in the unconnected mode,” she said. “It is already used to enable streaming on these TVs. It is not a big deal for device manufacturers.”

In the future, St. Jean added, they expect to begin offering multiple DRMs. “We are currently working with an affiliate group to figure out how to implement FairPlay for Apple devices,” he said.

St. Jean also said that there has been a lot of confusion around the issue of “signal-signing” certificates for 3.0 broadcasts. Signal signing is important, St. Jean and others explained, because it is designed to let viewers know that a signal is authentic and that hackers or pirates haven’t hijacked broadcast signals or emergency alerts. In many ways, the protection provided by signal-signing certificates is similar to the warning that consumers get online when they try to navigate to a suspicious website.

A3SA worked with Eonti and DigiCert to establish signal-signing certificates that will authenticate a broadcast signal. “We don’t make money from this,” he stressed. “All the revenue paid for the certificates goes to Eonti, who pays DigiCert. Our role is to work with broadcasters and device manufacturers to make sure they are able to accept signed signals and tell the difference between signed and unsigned signals.”

In response to complaints that those certificates were too expensive, Folliard said the industry has been working to provide more options and reduce costs. “Not everyone needs a full-service certificate for $950 [a year],” he said. “We’ve already cut that price down to $550, and we’re working to offer even additional lower-cost options.”

Folliard and others also explained that they hoped that more options would emerge as more companies begin offering 3.0-related security.

In November, for example, SiliconDust announced it is offering an Online Certificate Status Protocol solution, providing new security options and compliance to ensure TV stations meet all of the ATSC 3.0 standard’s signing-security requirements.

Schelle and Folliard praised the move. “We welcome all certificate providers,” Folliard said. “The more competition the better because that drives down prices.”

High Noon
Another major issue has been the so-called “high noon” for implementing signal signing. That deadline was originally scheduled for June 2025, but it has been indefinitely delayed.

“When I came on board at the A3SA, it was clear to me that the industry wasn’t ready to turn it on,” St. Jean said, because there wasn’t “consistent messaging on devices” and because “the consumer experience that you see out in the marketplace is a poor one.”

Currently, the timing for high noon is difficult to predict. Schelle and St. Jean said the initial decision to delay it was a collaborative process with the National Association of Broadcasters, the Consumer Technology Association, the Advanced Television Systems Committee (ATSC), A3SA and other stakeholders. They also stressed that any future deadline would come out of a similar collaborative process. “We won’t implement it until we have the right consumer experience,” St. Jean said.

Whether these changes will assuage the fears of LPTV stations and other smaller broadcasters remains an open question.

Kristina Bruni, the chair of the Advanced Television Broadcasting Alliance, said that while the group is still polling its LPTV station members regarding ATSC 3.0, she felt there was some confusion in the market about signal signing.

“I do not think that an LPTV operation that is going to transition to 3.0 is going to find that the signal-signing certificate will break the bank,” she said. “I feel that is a nonargument. Nobody is forcing them to go to 3.0 and if they do, the signal-signing certificate is also optional. It’s like when you are on a website and it doesn’t have the proper certificate. As a consumer, you can still choose to continue or not.”

Device Security
Another major question is the potential impact of content security and DRM on the ATSC 3.0 device market.

That issue exploded in 2025 in FCC filings related to SiliconDust’s HDHomeRun 3.0 receivers, which reportedly had problems decoding encrypted 3.0 signals. That prompted SiliconDust and thousands of consumers who own devices to argue that the FCC should not allow 3.0 broadcasts to be encrypted.

In response, Pearl TV told the FCC the problem was not encryption but that the fact that the HDHomeRun device uses a chipset manufactured by HiSilicon, a wholly-owned subsidiary of Huawei Technologies, a company “which the Commission has found to be ‘a national security threat.’”

No other manufacturer has had that problem, Foillard said.

SiliconDust “made a business decision to go with a Huawei chip that can’t support Widevine,” he said. “All the others will tell you supporting Widevine is not a big deal.”

SiliconDust has hotly disputed those claims, stressing that its devices are “NextGen TV-certified” and blaming A3SA’s implementation of Widevine for the problem. (Detailed coverage of this controversy can be found here.)

Other device manufacturers downplay the broader impact of the issue. “I don’t think it [security and DRM] has had a large impact on the device market,” said Alex Day, vice president of business development at Tolka Telecommunications, which provides a software platform that is used by ADTH and other manufacturers of 3.0 receivers.

Day expects lower-cost device options to come to market in 2026. Higher-end devices, gateways and gear with fully integrated Android TV capabilities should also launch, he said.

Gopal Miglani, president and founder of BitRouter, which aims to be the “TiVo of NextGen TV” with products like the ZapperBox receiver, said his company recently released its Unbound Gateway v3.4 software.

The software is “a significant milestone in bringing all the capabilities of the ATSC 1.0 DVR ecosystem to the ATSC 3.0 transition,” Miglani said in a blog post, with a multiroom DRM beta for gateway-connected devices. Software releases v3.5 and beyond will allow multiroom devices to control one another as real gateways, and 2026 will also see the creation of the first streaming app for a non-ZapperBox device.

Other industry players stressed the importance of the FCC setting a cutoff date for NextGen TV transition.

“There has not been a transition globally that has ever taken off until a date has been set,” Schelle said. The whole device market “really changes when the FCC sets a date.”

However, proposals for a 1.0 sunset face powerful opposition from the CTA and TV-set manufacturers. While the FCC has said it wants to speed the transition, the agency has not made any public moves to set firm dates or impose device mandates.

Throughout its century-plus existence, the motion picture industry has had to fight battles on multiple fronts to protect its content from piracy. From protecting film reels from duplication by unscrupulous parties (often theater owners themselves), to the age of VCRs, DVDs, IP-based peer-to-peer networks and today’s streaming services, Hollywood has a long history of working to keep ahead of what has often been termed as a “whack-a-mole” strategy.

And as content has moved from the theater to broadcast and now to streaming, the threats are myriad and growing, with the estimated annual lost revenue to the U.S. economy due to global online content piracy at nearly $30 billion, according to Bryan Willett, vice president, content protection (Americas), Motion Picture Association (MPA).

One of the biggest misconceptions is that piracy is a victimless crime and that it’s harmless, Willett said.

“In reality, digital piracy harms the entire entertainment ecosystem, including
all the people who work hard to bring a movie or TV series to fruition,” he said. “We’re talking about set builders, makeup and wardrobe artists, caterers, cinematographers, sound mixers, production assistants—the list goes on.”

In today’s connected world, that’s not the only risk, he adds.

“Streaming or downloading pirated content puts consumers’ sensitive information at risk,” Willett added. “Recent studies have shown that accessing pirated movies or TV shows could compromise credit cards, banking information and even infect your computer with malware.

“Many consumers aren’t aware that digital piracy operations are often linked to global organized crime rings that engage in drug and human trafficking, fraud and other crimes,” he added. “We like to say that if a streaming deal sounds too good to be true, it probably is.”

Strength in Numbers
The importance of intra-industry cooperation to battle the threats posed by piracy prompted MPA to create the Alliance for Creativity and Entertainment (ACE) in 2017.

“MPA/ACE have nearly 50-plus members, comprising studios, broadcasters and sports organizations, and we work with them regularly, particularly with respect to our anti-piracy efforts,” Willett said. “We also have developed relationships with many intermediaries and work with these and many other partners to ensure we continue to detect, deter and dismantle piracy operations.

“Working with these partners is critical, given piracy operators are reliant on them to provide infrastructure, bandwidth and payment processing,” he added. “Additionally, a great part of our success is due to the partnerships we’ve formed with law-enforcement agencies such as the FBI, HSI, Interpol and Europol, as well as other foreign agencies.”

The studios have used a variety of methods to combat piracy over the years, through physical enforcement, litigation and legislation, but perhaps the most significant step to address the problem in the digital age was the creation of MovieLabs in 2006.

Motion Picture Laboratories Inc. (aka MovieLabs) is a nonprofit technology research laboratory that was founded by the then-major studios, Paramount Pictures, Sony Pictures Entertainment, 20th Century Fox Film Corp., Universal City Studios, Walt Disney Pictures and Television and Warner Bros. Entertainment.

In essence, MovieLabs serves as a tech hub designed to research and fund the development of technologies that advance media distribution and creation, according to President and CEO Rich Berger, who noted the confluence of factors that has led to the challenges studios currently face.

“Throughout our history, our industry has faced a range of challenges, including changing consumer habits, rapidly evolving and disruptive technologies, more competition for time spent on media consumption and economic challenges,” he said. “Currently, we are seeing a combination of all these factors. The good news is that our industry continues to evolve and overcome these challenges.”

A key component of the organization’s current focus has been outlined in the “MovieLabs 2030 Vision,” which outlines its strategy in adapting to the rapidly evolving media distribution landscape.

“The MovieLabs 2030 Vision is a growing industry movement inspired by the MovieLabs Evolution of Media Creation white paper, which lays out a bold vision and roadmap for a more secure, efficient and interoperable media creation pipeline using cloud infrastructure, zero-trust security and software-defined workflows,” Berger said. “The ultimate goal is to empower storytellers to create the best stories imaginable without limits.”

Although battling piracy is among the core goals of MovieLabs, the organization is not an enforcer. Instead, it chooses to attack the problem by identifying and developing new technologies for content creation and distribution, which in turn, can result in the enhancement of copyright protection. Among its biggest current technical challenges is the inefficiency of moviemaking, according to Berger.

“One of the biggest challenges that MovieLabs and the 2030 Vision seeks to address is the current inefficiency and complexity of media creation,” Berger said. “While the industry has a number of challenges, including the general disruption that technology itself has created, the 2030 Vision can help meet today’s mandate for efficiency while maintaining the quality that audiences have come to expect from our studio members.”

Talking Together
Interoperability underpins just about every aspect of the 2030 Vision, according to Berger.

“One of the most important benefits of interoperability is the wider choice of vendors, tools, infrastructure and services it gives to creative teams,” he said. “And it has many others, including streamlining workflows to save time for higher-value creative tasks, enabling faster adoption of new, innovative technologies without breaking things, simplifying collaboration, and creating more secure and frictionless authentication and authorization experiences.

“For these reasons, it’s essential that our industry come together, as we are in the Industry forum, and make the new generation of workflows more interoperable,” he added.

To achieve this goal, MovieLabs has created a new group to promote such collaborations.

“We think the exchange of metadata that connects people, work and assets is ripe for improved interoperability, which led us to create the Ontology for Media Creation (OMC),” Berger said. “And we have an active program called “2030 Greenlight,” where we work with tool vendors to improve the interoperability of their products and to demonstrate their new capabilities, like the use of OMC, in real workflows.

“Ultimately, the 2030 Vision is about creative enablement,” Berger added. “2030 Greenlight is all about demonstrating the value for our creative partners.”

Trusted Partner Network
Another organization created in recent years to take on piracy is the Trusted Partner Network, a joint venture between MPA and the Content Delivery & Security Association (CDSA) formed in 2018. Among its first goals was to use MPA content security best practices to assess security risks in the industry, TPN President Terri Davies said.

“At the end of the first year [2019], they had something like 300 assessments completed,” she said. “All of the studios were signed up, many of the service providers in the industry were signed up, and it was going really well.”

Then the pandemic exposed a weak point.

“The best practices and TPN were heavily focused on physical site security; they hadn’t yet moved to app and cloud—and of course, we know what happened,” Davies said. “COVID closed all of the buildings and the security assessments meant nothing.

“Everybody was very, very entrepreneurial at that time, and those who could move to the cloud did move to the cloud, and TPN didn’t keep up,” Davies added. “And the studios had no choice then but to go back to this very fragmented approach of doing their own security assessments of each of the service providers, because the real goal of TPN is to improve security preparedness in the industry but also to provide efficiency.”

In 2023, the organization was relaunched with a new membership model, application and cloud content security assessments, and the introduction of the “TPN+” platform.

The organization—which is now wholly owned by the MPA—used the updated MPA Content Security Best Practices v5.0 to expand its scope to include software application and cloud security in addition to site and work-from-home security assessments.

“With the launch of its enhanced program, TPN is taking a crucial step toward ensuring the security of all content produced by our members,” Karyn Temple, senior executive vice president and global general counsel of the MPA, said. “Content security is a top priority of the MPA, TPN and all of our member studios, whose ability to produce the films and TV shows audiences love depends on our ability to protect and secure that content.”

Davies explained the three main roles the TPN provides to the industry.

“We have three roles in TPN—one is what we call a content owner, which might be a Hollywood studio, Sky or BBC studios,” she said. “Those are the content owners who consume the security assessments from the vendors, who we call service providers, who do the security assessment in TPN. The service providers are everything from visual effects, localization, even pre-script.”

Davies estimates that 90% of TPN’s members work on pre-release content, “so we have a lot of application providers, like Adobe, frame.io, for example, as well as the traditional service providers—editing, localization, audio and post, all the way through to distribution.”

TPN’s third role is the aforementioned risk assessment, Davies added.

“We have our own accredited assessors who do the assessments, because we can’t just have anybody do our assessments,” she said. “We consider all three of those roles and anybody—whether they paid us or not—that’s in the platform, they are our stakeholders. We send out security alerts to them, for example, when we hear something going on through ACE, or through the studios or through the industry.”

Membership in TPN is voluntary but rapidly growing, Davies said. It includes all seven MPA studios, along with new members Apple Studios, BBC Studios, Sky and Canal Plus Group.

“We’ve got over 1,500 companies signed up in the TPN+ platform right now representing 60-plus countries,” Davies said, adding that “78% of our members have less than $5 million annual gross revenue.”

All of those members have a vested interest in industry-wide content protection and with the three organizations collaborating, they are making a difference, according to Willett.

“ACE takes down numerous piracy operations each year, he said. “In fact, a few months ago we shut down the world’s largest live sports-piracy network, Streameast. The site received over 1.5 billion visits in the past 12 months, representing major financial losses for our live-tier members.

“Each time a court sentences a piracy network operator for their illicit activities, we consider that a victory,” Willett added. “It sends a powerful deterrent message that piracy is a crime and the consequences are serious. We’ve seen judges issue fines and even prison sentences for piracy operations of all sizes.”

As the malevolent siege against broadcasters’ interests intensifies from the far reaches of artificial intelligence misuse to relentless innovation in the multibillion-dollar piracy business, the overarching question is whether the industry can muster resistance on a scale conducive to success.

So far, there’s much to applaud on the anti-piracy side of the battle. When it comes to thwarting the destructive use of AI, though, the pace of remedial action is lagging well behind the growing threat, despite the din of alarms sounding from all corners of the marketplace and a plethora of federal and state legislative initiatives.

“The rise of large-scale content scraping by AI systems and digital platforms poses a serious threat to the sustainability of local journalism,” Alex Siciliano, senior vice president of communications at the National Association of Broadcasters, said. “These systems use broadcasters’ reporting to train or populate AI products without authorization, compensation or attribution—diverting revenue away from local stations that invest heavily in producing trusted news, sports and emergency coverage.”

Whether Washington has the will to do anything about this set of AI issues remains to be seen. The latest addition to the stack of pending Senate bills—the AI Accountability and Personal Data Protection Act, introduced in late July—awaits any sign that it will garner consideration.

On another track, the Content Origin Protection and Integrity from Edited and Deepfaked Media (COPIED) Act, which stalled in the Senate last year, has been reintroduced with full support from NAB to create barriers to use of deepfakes and unauthorized use of content to train AI models. Another of those NAB-supported efforts from 2024, the NO FAKES Act, was reintroduced this year with broad bipartisan support in the House and Senate. That measure takes a complementary approach toward, as NAB puts it, “protecting trusted broadcast journalists, local radio hosts and other on-air personalities from the unauthorized use of their voice, image or likeness.”

Obviously, moving all these different measures to passage, possibly with efforts to consolidate them at some point, will take a good deal of time. And there’s no telling what will be done to deal with the regulatory chaos stemming from state-level initiatives, which to date have produced over 1,000 bills targeting AI regulation of some kind, according to the security regulations tracker Lawfare.

‘Scary Good’ AI Apps
Meanwhile, just how far ahead of the law-making curve Big Tech support for mass acculturation to AI hyperrealism has gone is seen in the success of OpenAI’s Sora app, dubbed “scary good” in a New York Times headline for its ability to act as a TikTok-like social media engine for sharing AI-generated videos. In fact, as TV Tech reported in July, there’s even some experimentation with AI fakery on the part of station groups that want to bolster the reach of their newscasts by taking advantage of its ability to distort reality by turning monolinguistic on-air personalities into fluent multilinguists.

Such steps, taken with public notification of AI usage, are a response to pressures arising from the need to build broadcast news viewership in the digital domain. While TV news still serves vast audiences, “it doesn’t have the relevance, the scale and necessarily the best business model when it gets into the digital space,” said Michael Newman, director of transformation for Graham Media Group’s digital operations. At this point, Newman said, using AI in on-air newscasting is a bridge too far, though he acknowledged is use to support live translation in newscasts is “really cool and a great example of how AI and original reporting can be accessed by more people.”

Even though broadcasters welcome some controls—as reflected in the NAB’s support for legislation like the COPIED Act—the way forward is complicated by inevitable disputes over how to rein AI in. “We are also engaged with policymakers and industry partners to ensure any new regulations or voluntary standards maintain a balance between protecting creative rights and allowing legitimate innovation,” Siciliano said.

Inevitably, as the slow grind unfolds in Congress and at the Federal Communications Commission—which, at press time, was scheduled to vote in late October on a Biden-era notice of proposed rulemaking governing disclosure of AI usage in political ads opposed by Chair Brendan Carr as well as the NAB—it remains up to stations to set up protective procedures against AI misuse by outsiders as well as parameters defining limits on internal usage. To many in an industry where attitudes about regulation are mixed, this isn’t necessarily a reprehensible situation.

At Fox Television Stations, “the approach overall is that this is an area of growth that’s happening that we need to understand and be involved in,” said Jeff Zellmer, the group’s executive vice president of digital operations. “It’s better to find partners that help us use it well than to throw up walls.”

Added Sinclair Chief Innovation Officer Scott Ehrlich, “Moving responsibly falls to companies like ours.” As for industry-wide agreement on such things, “it’s hard to standardize when you’re dealing with a target moving as fast as AI,” he said.

Piracy Concerns, Too
Nonetheless, while the industry hasn’t formalized standards around AI usage, Ehrlich voiced hope for organizations moving the industry toward more generalized protection that encompasses piracy as well as AI. One group he cited is the Coalition for Content Provenance and Authenticity (C2PA), an ad hoc group working to systematize ways to establish content provenance and address misinformation.

Indeed, it’s in this realm of tools engineered to “help broadcasters verify the origin of their content, signal to viewers that programming is genuine and trace misuse when it occurs,” as described by Siciliano, that progress on the anti-piracy front offers a hopeful counterpoint to the slog on the AI side. Broadcasters are “collaborating with technology partners to strengthen the resilience of their distribution systems and guard against unauthorized access or signal manipulation,” he noted.

At this point in time, piracy is far more damaging to bottom lines across the TV broadcasting and streaming ecosystem than AI. “Video piracy is the third largest category in organized crime behind drugs and sex trafficking,” said Robin Boldon, head of product at Friend MTS, a leading supplier of forensic fingerprinting and watermarking tools used, respectively, to find stolen content in the global stream flow and identify its sources.

A 2024 survey conducted by streaming service tracker cordcutting.com found that one in three U.S. adults admitted to pirating TV shows or movies within the past year. According to research reported by the Kearney consulting group, the U.S. ranks first in national tabulations of visits to pirate websites. Globally, the number of visits to such sites rose by 13% from 125 billion in 2019 to 141 billion in 2023, Kearney said. Parks Associates has predicted U.S.-based content suppliers will lose $113 billion to piracy in this country alone by 2027.

Big Anti-Piracy Advances
It’s clear that a new generation of advanced anti-piracy tools and services is making life harder than ever for thieves. Silent Push, operator of an advanced piracy scanning and aggregation engine, recently reported uncovering an IPTV piracy network impacting more than 20 global media brands through operations involving over 10,000 IP addresses identifying servers and computers associated with about 1,000 websites.

Friend MTS has expanded its support for fighting piracy on several fronts. In response to the recent surge in so-called CDN leeching involving injection of purloined digital rights management keys, the company launched its server-side “Distribution Watermarking” solution to identify hacked subscriptions restreamed by pirates. This complements the widely deployed FMTS client-side Subscriber Watermarking solution that disrupts a live stream within a few minutes of start time, which is essential to causing meaningful pain to viewers of illicit sports feeds.

Most recently, FMTS upgraded its piracy monitoring service with use of an AI Retrieval-Augmented Generation (RAG) model that continuously learns and adapts to pirates’ evolving tactics. It also launched fully staffed Piracy Investigation Services, which put regional teams to work using FMTS tools in ways that combine threat intelligence, identification and enforcement in accord with local practices.

“Along with supporting traditional methods like sending notices and building the chains of evidence required by courts, we’re going many steps further with service layers offering investigations and enforcement assistance, depending on what customers want,” Friend’s Boldon said.

Shift in Attitude
Another adjustment to the current state of affairs comes from Nagravision, a longtime leader in traditional conditional access, multi-DRM protection, watermarking solutions and on-the-ground teamwork with local law enforcement and regional organizations. The company has introduced a new approach to shaping anti-piracy strategies with a business-first perspective on what customers need, said Tim Pearson, vice president of global solutions and partner marketing.

In a market characterized by the diversity of issues associated with the transition to streaming, “we’re seeing buyers coming with very purposed agendas,” Pearson noted. “They have a problem, and they want to fix it.” He said the company is putting AI to work perusing “massive amounts of analytics and insights into data platforms” to find “pattern matches that allow us to really understand consumer behavior,” such as discerning how service price deductions can produce net ROI gains by reducing reliance on pirate services.

Nagravision is also teaming with Broadpeak to strengthen the battle against CDN leeching in use cases involving live sports. With AI abetting analytics, the solution can monitor DRM license requests to detect suspicious activity patterns and speed the disruption of illicit streams, Damien Sterkers, vice president of products and solutions marketing at Broadpeak, said. “On top of the video intelligence layer, we have human services with people working 24/7 to inform content owners what’s happening,” he said.

Underlying the effectiveness of all this activity is a big shift in content owners’ attitudes about piracy. “When piracy was detected, it used to be seen as the rights holder’s problem,” Boldon said. “Now it’s understood that it’s everybody’s problem.”

A case in point is growing participation in the International Broadcaster Coalition Against Piracy (IBCAP), which has expanded to include international and U.S. content owners, broadcasters and distributors representing more than 220 television channels. IBCAP said it has instigated a surge of lawsuits running into the tens of millions of dollars against wrongdoers in the U.S. and abroad, along with real-time takedowns of illicit live sports streams, including disruptions of more than 6,000 streams reaching millions of viewers during the 2024 Indian Premier League cricket tournament.

Whether cooperation in battling the misuse of AI can get to this level remains to be seen.

With the total value of the global digital video market estimated at $193 billion in 2023 and expected to explode to more than $503 billion by 2032—according to IMARC Group—that growth rate of nearly 11% represents a market that needs to make content protection a top priority.

“The growth trajectory means it will be a dereliction of responsibility for companies in the digital content space that derogate or underfund efforts to protect their investments and IP,” said Emeka Okoli, senior vice president of business solutions with Zixi, a secure content streaming platform provider for companies such as Amazon, Bloomberg, Apple, Fubo and Roku. “All linearized content, including big-screen VOD assets, are fair game to the bad guys who also see a gold mine.”

In the past, broadcasters and content producers could live with a certain level of piracy and look the other way, according to Asaf Ashkenazi, CEO of Verimatrix; developer of streaming security solutions such as Verimatrix Streamkeeper, whose Counterspy module uses AI to anticipate attacks. “However, today’s incredibly competitive market forces everyone to be more efficient and make sure they don’t share their value with cybercriminals and pirates.”

A Wide Range of Threats
Given the multibillion-dollar value of digital video content, it’s not surprising that it’s under attack from many directions. “Key threats include piracy, content leakage, theft and illegal distribution, all of which compromise content monetization,” said Ian Hamilton, CTO of Signiant, known for its SaaS solutions that support secure content exchange within and between media organizations. “Piracy and illegal distribution result in lost revenue from unauthorized access, while content leakage and theft devalue exclusive releases.”

There is a mind-boggling variety of threat actors doing their best to steal content and then profit from that theft, according to Hamilton. “Threat actors range from organized crime groups and hackers to insiders, individual pirates, and even state actors,” he said. “Organized crime groups run piracy operations for profit, distributing stolen content through illegal channels. Hackers may target content for financial gain or to disrupt media companies. Insiders, such as employees or contractors, might leak content for personal or financial reasons. Individual pirates steal content for notoriety or minor profits, while state actors may pursue political or economic objectives.”

To make matters worse, today’s video content pirates are not the stereotypical lone hackers sitting in a darkened theater with a camcorder, or recording video streams on their home computer and then sharing those copies with friends. In many cases, “they are organized just like a profitable digital streaming business, often operating across borders with sophisticated operations centered around one goal: making money at the expense of hard-working content providers and video distributors,” said Ashkenazi.

The Pros and Cons of AI
The advent of AI isn’t helping. “AI enables threat actors to automate attacks, enhance social engineering, and create deepfakes,” said Hamilton. “It facilitates rapid identification of vulnerabilities and streamlines the theft and distribution process.”

The bottom line: “Anyone who can steal content probably will,” observed Ben Jones, director of solutions marketing at MediaKind. “Their motivations include profit from selling pirated content or bypassing paywalls to watch stolen content for their own personal gain.”

Advances in technology are aiding thieves in their attempts to steal digital video content, but technology is also helping content producers and distributors to better protect themselves against such theft. This technology is being harnessed by companies like Akamai, MediaKind, Signiant, Verimatrix and Zixi, to deter and disable such attacks when they occur.

At MediaKind, their clients are increasing their use of digital watermarking—hiding unique, traceable identifier data within the content, so that owners of pirated content can be found and alerted. “At any point in the chain where there’s been a detection of leaked content, we can work out where that occurred using watermarking,” Jones said. “With this data, content owners can point their lawyers at the pirates, and force them to take the content down.”

Upgrading Encryption
Increasing the complexity of cipher blocks and secure algorithms is also making it harder for thieves to steal content digitally. “The older encryption system used a 42-bit key; now we are moving into 228- and 256-bit keys, which are much more military-style and secure,” said Jones. “We’re also using measures such as DRM [digital rights management] and real-time content monitoring online.”

Zixi’s SDVP and telemetry plane, Zen Master, includes high AES encryption. “We also provide DRM functionalities that utilize DTLS [Datagram Transport Layer Security], which safeguards against stream ripping, eavesdropping, man-in-the-middle attacks and fraudulent endpoints by verifying the receiving device’s certificate of authenticity,” Okoli said.

“From an Akamai lens, since we’re dealing mainly with streaming services delivering video content to end users, primary threats to content include theft, piracy, rebroadcasting, automated scraping and account hijacking,” noted Tedd Smith, solutions engineering manager at Akamai Technologies. “These types of actions can be incredibly damaging, causing everything from decreases in viewership, loss of content rights, brand damage, increase in enforcement costs, and revenue loss.”

“Furthermore, Zixi’s platform includes DPDK (Data Plane Development Kit) high-performance networking, enabling content to operate at the network layer, bypassing the operating system and effectively blocking potential OS-based ransom malware. With Zixi’s new multi-hop product, we can offer customers upstream metadata observability which will help companies proactively trace their content incoming digital journey and footprint.”

Signiant’s content protection protocol employs integrated multilayered strategies to protect its clients. “This includes using AI for real-time monitoring and threat detection, applying strict access controls based on least privilege, and employing encryption,” said Hamilton. As well, “Minimizing copies of content by interfacing with customer-owned storage reduces exposure and potential vulnerabilities.”

Akamai and Verimatrix also use the many content protection strategies described above.

Constant vigilance is another important tool: “Visibility is a key component of content protection as well, especially when dealing with direct-to-consumer delivery,” Smith said. “This is critical during live events when immediate mitigation is a top priority. Having real-time visibility into traffic patterns and user behavior allows streaming services to quickly catch things like stream-sharing and rebroadcasting so access can be revoked.”

Overall, “our commitment is to make piracy unprofitable, illegal distribution unsustainable, and content security an enabler of business growth rather than merely a compliance checkbox,” said Ashkenazi.

Verimatrix’s position is shared by the content protection companies interviewed for this story, all of which want to ensure that crime does not pay for pirates, and that legitimate video producers/distributors continue to profit from the multibillion-dollar value of their content.

The media and entertainment business is built on what it offers its viewers: programming, sports, news and other live events. While language purists may balk at the current use of the word “assets” instead of “programs,” it does convey the value of the material—both creative and commercial—shown on TV channels and streaming services. They are valuable commodities and have always needed to be protected. Unfortunately, the move to file-based production, with everything on computer servers and network or cloud access, has left assets—and their owners—extremely vulnerable.

This was starkly illustrated by the 2014 hack of files and data from Sony Pictures, which included staff details, emails, corporate information and copies of at-the-time unreleased films. Calls for heightened security in the aftermath of the breach ultimately led to the creation in 2018 of the Trusted Partner Network (TPN) by the Motion Picture Association (MPA) and the Content Delivery and Security Association (CDSA), with the MPA taking full control of TPN in 2021.

Early Moves
The MPA issued its first best practice guidelines for content security in 2009, in collaboration with member companies and consultancy Deloitte. Updates and revisions followed between 2011 and 2018, with authorship moving exclusively to the MPA and its membership.

Deloitte published its own report on the subject in 2018, highlighting the digital transformation of content in all its forms. Among its recommendations for securing material were digital fingerprinting, encryption, blockchain (a digital ledger for recording transactions and events), watermarking and plugging the “analog hole,” where programs are copied from legacy sources.

John Footen, managing director of Deloitte Consulting, acknowledges that the situation continues to evolve but says new technologies are now being used with established practices to further tighten security.

“Automation can reduce the effort to maintain a consistent security posture in large-scale systems and to recover after an attack,” he explains. “We’re also seeing greater integration of media technology with enterprise security systems. More products can work with single sign-on in zero-trust environments, for example. At Deloitte, we’re applying AI for cyber defense, not only for well-established uses such as detecting anomalous behavior and interdicting threats but also for assessment and visualization.”

The changing nature of the threat— hackers developing new viruses and ways of infiltrating systems—calls for content owners to be ever vigilant. This is reflected in the MPA/TPN Best Practice Guidelines, which have been updated on an annual basis (sometimes twice in a year) since 2018. The most recent revision appeared in August this year, with new guidance on working from home and remote access to data centers, plus minor updates to cloud specific controls.

The recommendations from Crystal Pham, vice president of operations and product management for TPN are fairly universal no matter the process: Ensuring that content owners and facilities have secure connections— including not logging into applications from unsecured networks; keeping applications updated and patched; using strong passwords and authentication mechanisms; not sharing accounts or credentials; always installing applications from trusted sources; and adhering to MPA Best Practices, as well as completing an annual TPN assessment.

Pham characterizes the security issue in M&E as “always evolving” and can be influenced by new technologies, the shortening of  release windows and emerging threats. 

“Change is a constant,” she says. “I do think there’s more awareness about security, with better understanding of the constant change and the need to get ahead of it when possible. I believe it’s becoming more embedded in the process of creating, distributing and consuming content. It’s also becoming more ingrained in our culture.”

New Methods Needed
The ever-shifting landscape is summed up by Asaf Ashkenazi, chief executive of Verimatrix, who observes that as technology advances, so do the techniques used by pirates to illegally obtain, distribute and monetize content.

“DRM and other traditional security methods are no longer as effective against today’s savvy hackers,” he says. “New real-time detection and prevention systems that are layered on traditional protection methods are needed to combat modern piracy efforts—especially those that leverage AI/ML and cloud infrastructure to exploit at scale.”

To counter such threats, Verimatrix has developed Streamkeeper, which combines multi-DRM, forensic watermarking and “Counterspy,” which Ashkenazi describes as a “motion sensor” or “alarm” that detects piracy as it happens. “The goal, however, should not be to completely eliminate piracy but to disrupt the pirates’ business models,” he comments. “By increasing costs beyond potential profits, the motivation for piracy diminishes. Real-time detection, prevention and rapid response are key.”

Among the ways of disrupting the activities of pirates is to track down purloined programming to the sites where it is being offered illegally and shut them down. Another company providing such monitoring enforcement services is Friend MTS, which does this using both fingerprinting and watermarking technologies. 

“A customer will give us a clean reference feed of their playout and we fingerprint that,” explains Nik Forman, vice president of marketing for the U.K.-based company. “This allows us to uniquely identify that clip of video. If the monitoring systems— which are all automated and crawling the web—then find a suspicious web site is playing that content, we fingerprint it so we can compare them. 

“If they match, we know it is an illicit player and the client can send an enforcement notice or a takedown order,” Forman continues. “If they also take our watermarking services, the content is marked invisibly and basically sits like a QR code in the actual video. When we find the illicit content and we identify it, we match it with a fingerprint and can then say it is one of our watermarks and we extract that payload of subscriber information.”

‘Never Trust, Always Verify’
While the Sony hack made the M&E sector painfully aware of its vulnerabilities, the threats have only increased since then due to otherwise innocent technological developments. The threat landscape has evolved with emergent technologies such as the Internet of Things, 5G connectivity and advanced generative AI capabilities, according to Eric Elbaz, principal strategic engagement manager at Akamai.

“We’ve also seen state-sponsored attacks become more brazen,” he said. “Nation-state actors are believed to be behind some of the most sophisticated and persistent threats seen to date.”

This was the suspicion surrounding the attack on the supply chain of software company SolarWinds in 2020, while, as Elbaz points out, ransomware attacks have increased “dramatically.”

A recent example of this is the breach of MGM Resorts’ systems in Las Vegas, which reportedly cost the company approximately $100 million. But Elbaz does see progress in dealing with the threats: “On the defensive side, we’ve seen major strides being made across threat intelligence, a paradigm shift from ‘trust but verify’ to ‘never trust always verify’ architectures, which is zero-trust security. Deeper public-private sector collaboration and information sharing, as well as a catch up in privacy laws, regulations, and standards are also helping.”

The threat shows no signs of going away and future attacks are likely to be even more cunning and sophisticated. According to “Variety,” in 2022, hacking of popular entertainment content rose 18% YOY compared with 2021, with the U.S. having the largest share of film and TV demand (i.e., illicit streams, downloads and the like), with more than 13.5 billion visits to piracy sites.

But the tools to combat it are there and, when it comes down to it, the greatest weapons are probably vigilance and common sense. 

CAMBRIDGE, Mass.—Akamai has announced a range of new steps aimed at protecting streamed broadcast content, including a provision giving rights holders the ability to execute immediate shutdowns of unauthorized streams or the source of piracy. 

In a blog posting, Akamai’s Chris Nicholson laid out the content protection strategy, including the immediate stream shutdowns. 

“Access Revocation can work on its own or integrated with either the Friend MTS or NAGRA watermarking solutions to be part of a formidable ‘prevent-discover-enforce’ content security posture,” wrote Nicholson.

Immediate halts to illegal streams are particularly important for live sports where the rights holder’s ability to make the most money from the event being covered normally is greatest during “a narrow window of competition,” he said.

Nicholson quoted from an interview by SportsPro with BeIN Media Group Chief Executive Yousef Al-Obaidly underscoring the urgency with which pirate sites must be shut down, particularly when it comes to sports. “You either make your money now or you don’t make money,” the blog quotes him as saying in the interview.

According to the blog, the company has taken other steps to fight piracy, including protecting against credential theft, system hacking, rebroadcasting, stolen streams and VPN abuse. Akamai has also signed on various partners with unique watermarking solutions to its content security ecosystem. They include:

• ContentArmor forensic watermarking; 
• Friend MTS ASiD watermarking; 
• INKA Entworks PallyCon forensic watermarking; 
• Irdeto TraceMark; 
• NAGRA NexGuard forensic watermarking; and 
• Verimatrix watermarking, StreamMark version 

The same ecosystem also has the ability to protect linear/simulcast streaming and video on demand, Nicholson added. 

LAS VEGAS—At the 2019 NAB Show, MediaKind launched Cygnus Director 128, a control and conditional access management software product. The newest generation of MediaKind’s Cygnus Primary Distribution platform, Cygnus Director 128 enables broadcasters and content owners to secure and manage the primary distribution of their content to secondary broadcast operators, by identifying and helping to prohibit the unauthorized streaming and broadcasting of their live and on-demand content to consumers.

MediaKind’s Cygnus Primary Distribution solution utilizes advanced 128-bit encryption coupled with forensic watermarking technology to deter and help prevent piracy of high value content from the broadcaster/content owner’s distribution network, as well as better protecting subsequent copies or streams.

Coupled with the newly enhanced encryption, Cygnus Director 128 provides a management platform for further control, upgrade and standards alignment with the following benefits:

• Individual and group management of remote professional receivers including entitlements, forced tuning to avoid satellite sun outages, and forced channel selection for automated blackout management, and over-air software download for network wide updates.
• The advanced reporting system will give broadcasters and content owners new operational visibility on the configuration status of their entire network of professional receivers.
• Offers the ability to deter, detect and prohibit piracy of high value content, coupled with control over who can view content, including affiliate networks that wish to receive and access their high value content.
• Cygnus Primary Distribution security solution can be applied over either an IP, Satellite or hybrid Primary Distribution environment, and will also support the BISS2 conditional access to ensure support for open industry based standards.
• The current global deployed list of Cygnus Director customers will be able to easily upgrade to the newly enhanced security from their existing systems utilizing an IP return path.