United States - Market Insights (Media)

Synthetic content, real consequences: AI risk in the media sector

The operating reality

Most of the public conversation about AI in the media still fixates on generative content. AI-written scripts, cloned voices and generated images raise obvious questions about rights, consent and the potential for brand damage. But inside media companies, AI is already doing quieter work. AI is already being used operationally, including in marketing tools, archive tagging, recommendation engines, captioning, localization, ad targeting and moderation. Different teams often adopt different tools, with little communication or standardization. Tools are frequently rolled out at different times and governed, if at all, in different ways.

That fragmented adoption is where legal risk hides. The question is not whether a company uses AI. The question is whether the company understands the role of AI in its workflows and what data and content move through it. Is the AI an internal company developed system, or is it an external system controlled by a vendor? Can the company explain how the AI output was created and what it was trained on?

One practical consequence is that AI oversight cannot be limited to obvious content generation tools. The larger exposure may sit in metadata, ranking, recommendations, moderation, ad targeting and vendor tooling. Those parts of the business rarely get treated as publishing until something goes wrong.

AI as infrastructure, not just content generation

The highest-value AI deployments in media may be less visible than synthetic content. LatentView Analytics describes high-maturity media use cases as including metadata enrichment, content operations, recommendations and churn prediction, and emphasizes that AI spans content creation, operations, distribution, monetization, trust and safety, and rights integrity (LatentView Analytics, AI in Media: Use Cases, Risks, and Strategic Implications for Media Organizations, updated March 4, 2026). That framing is useful because it moves the discussion away from the narrow question of whether a chatbot wrote a story and toward the practical question of how AI changes the company’s operating model.

AI tools now assist with transcription, captioning, translation, image recognition, quality control, archive management, clipping, content packaging and rapid distribution. Veritone describes AI systems that analyze large volumes of unstructured audio and video, automate processing such as auto-tagging, and turn media libraries into structured, searchable data (Veritone, Transforming the Future of Media with AI, February 12, 2026). This can make archives more useful and commercially valuable. It can also create questions about consent, biometric information, rights metadata, retention, reuse and whether the company can audit the AI’s decision.

AI can be useful in these areas, as manual work can be reduced and accessibility can be improved. Companies can also extract value from content that may otherwise sit unused. Operational AI is often treated as lower risk because it is not creating a published article or synthetic performance. While it may have less visible risk, there are still legal issues that should be reviewed. Inputs, rights in the underlying assets, audience impact, the data flows, and the company’s ability to review or reverse the output can all come into play. A recommendation engine, rights tool or ad classifier can create legal and business consequences even when they are not thought of as content creation.

Rights, ownership and synthetic identity

The most visible legal concern remains intellectual property. Media companies are often both AI users and rights holders whose works may be used by others to train AI systems. That dual role creates tension. A company may want to deploy generative tools internally while also objecting to unlicensed use of its own content.

In the United States, the Copyright Office’s 2025 report on copyrightability states that existing copyright principles can address AI-generated outputs, that human creativity remains central, and that purely AI-generated material or material lacking sufficient human control is not protected by copyright (US Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability, released January 29, 2025, www.copyright.gov/newsnet/2025/1060.html). This is consistent with the recent denial of certiorari by the US Supreme Court in Thaler v. Perlmutter, 130 F.4th 1039 (D.C. Cir. 2025), certiorari denied, No. 25-449 (US March 2, 2026), which left undisturbed the refusal to register a work identified as created autonomously by an AI system which was listed as the sole author on the copyright application. For media companies, that position is commercially important. If AI-assisted content contains meaningful human authorship, it may be protectable. If the output is essentially machine-generated, it generally is not protectable, meaning that the company will likely have no control over any future uses, edits or derivative works.

Training data is less settled. On May 9, 2025, the Copyright Office released a pre-publication version of Copyright and Artificial Intelligence, Part 3: Generative AI Training, addressing the use of copyrighted works in training generative AI models (US Copyright Office, Copyright and Artificial Intelligence, Part 3: Generative AI Training (pre-publication version), May 2025, www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-3-Generative-AI-Training-Report-Pre-Publication-Version.pdf). The Copyright Office treated fair use as a fact-specific inquiry and did not resolve any of the issues with respect to the legality of any particular model or training dataset. Litigation continues to develop in the space, with numerous cases working their way through the court system.

Entertainment and talent issues add another layer. Synthetic identity can include cloned voices, digital likenesses or AI-assisted performances that may require careful attention to existing laws and contracts, as well as guild or union regulations. California’s AB 2602 addresses contracts involving the creation or use of digital replicas of performers, including disclosure and/or representation requirements, and AB 1836 addresses unauthorized digital replicas of deceased performers. These laws focus on clear disclosure, permission, scope, and downstream control. Media companies should already be familiar with such issues, as they already come into play with real voice, likeness and performance usage.

The trust problem

The commercial risk of AI in media is not limited to ownership or infringement. Many media companies trade on trust. Audiences may not distinguish between a copyright infringement and poor editorial judgment. But both can cause the company harm.

“AI slop” describes generic, low-effort AI output. Such content can read as interchangeable and has the ability to weaken a brand’s voice. The risk is that repetitive AI-generated material can push audiences away and weaken differentiation (AlphaSense, Nicole Sheynin, AI in the Media Industry: Key Trends for 2026, March 13, 2026, www.alpha-sense.com/resources/research-articles/ai-media-industry/). There is no per se legal issue involved, but it is a real business constraint. A publisher can stay within the law and still erode the trust and distinctiveness that make its content valuable.

Hallucinations present a more familiar legal concern. If AI-generated or AI-assisted content includes false factual assertions about a person or business, publishers might not be able to avoid responsibility or liability by pointing to usage of an AI tool. The same concern applies to summaries, headlines, push notifications, captions, and search snippets.

Academic commentary points to the same problem. Runyan Wang’s 2025 paper, Artificial Intelligence and Change in the Media Industry: Opportunities, Challenges, and Ethical Considerations, identifies misinformation, algorithmic bias and unresolved ownership issues as recurring risks as AI expands from recommendations into content generation, audience engagement and media management (Runyan Wang, Artificial Intelligence and Change in the Media Industry: Opportunities, Challenges, and Ethical Considerations; Proceedings of the 4th International Conference on Literature, Language, and Culture Development (2025), DOI: 10.54254/2753-7064/60/2025.22842). Scale is what changes the exposure. Human editors make mistakes, but AI systems can make mistakes at volume.

Personalization and platform dependence

AI-powered personalization is now central to the media sector. Recommendation engines shape what users see. Predictive analytics inform what a user may want next. Advertising tools optimize audiences, pricing, placement, and creative variations. These systems can create value because they often turn data into engagement which can then turn engagement into revenue.

They also create legal exposure. Data-driven personalization depends on personal information. That raises privacy, consumer protection, discrimination and transparency questions. In September 2024, the Federal Trade Commission announced Operation AI Comply warning companies to keep AI claims in check and emphasizing that products marketed as AI-powered must work as advertised (FTC Announces Crackdown on Deceptive AI Claims and Schemes, September 25, 2024, www.ftc.gov/news-events/news/press-releases/2024/09/ftc-announces-crackdown-deceptive-ai-claims-schemes). The FTC’s immediate targets were not media companies, but the principle can be applied, nonetheless. AI-powered claims, audience products, branded content, and personalization features should work as represented and not mislead users.

Platform dependence is another significant concern. AI-powered search overviews and other “zero-click” experiences can answer user questions without a visit to the publisher’s site, shifting leverage toward the companies that control models and distribution layers, instead of the content layers (AlphaSense, Nicole Sheynin, AI in the Media Industry: Key Trends for 2026, March 13, 2026, www.alpha-sense.com/resources/research-articles/ai-media-industry/). This traffic disintermediation from AI-powered search overviews and zero-click results threatens publishers’ direct traffic and can therefore have a direct effect on advertising monetization (Ibid.).

This is not only a business development problem. If AI systems controlled by third parties reduce traffic, summarize content, demote real or licensed material, misclassify advertising inventory or change distribution rules, media companies may need to examine their contracts and regulatory obligations. Companies should also understand their options for challenging or responding to such third-party actions.

The workforce issue is also a legal issue

The effect of AI on labor and talent is an especially sensitive issue in media and entertainment. Some functions are being supplemented by AI, while others are being redefined or compressed. For example, AI may create new AI-related roles while also affecting creative and technical functions, including 3D modeling, character design, sound editing, music editing and scriptwriting.

Workforce changes tied to AI can affect more than just staffing. The media sector also needs to be cognizant of a host of additional issues, such as guild obligations, union negotiations, credit, residuals, consent, confidentiality, publicity rights and moral rights. These changes may also affect culture. A company that describes AI as a simple efficiency tool, while employees experience it as a threat to creative control, will have an internal credibility problem long before it has any potential legal one.

The language used to justify AI adoption matters. Efficiency alone is rarely enough if AI changes who performs creative work, how that work is credited, or whether talent has consented to reuse of their name, image or likeness.

Comparative regulation: The United States and the EU

The United States does not currently have a single comprehensive media-specific AI statute or regulation. Instead, risk is being addressed through federal copyright law, advertising and communications regulations, federal and state consumer protection laws, and a patchwork of state AI, contract, rights of publicity, digital replica, privacy and biometric laws, labor rules, and platform regulation. That makes the US environment flexible but fragmented. Media companies may face different legal requirements depending on whether the issue involves copyrightability or training data, or synthetic likenesses or advertising claims. State-specific laws may also come into play where the AI system or output uses personal information and biometric identifiers or involves employment-related decisions.

The European approach is more structured. Article 50 of the EU AI Act imposes transparency obligations for certain AI systems, including disclosure requirements for synthetic and AI-generated or manipulated audio, image, video or text content. Article 27 of the Digital Services Act requires online platforms that use recommender systems to explain in their terms and conditions (in intelligible language) the main parameters used in those systems. Article 18 of the European Media Freedom Act addresses the relationship between media service providers and very large online platforms, including declarations concerning editorial independence and human review or editorial control of AI-generated content.

For US-based media companies with international audiences, this comparative landscape can be difficult to navigate. A domestic policy that works for one US use case may not satisfy European expectations around transparency, recommender systems or platform accountability. The key is to anticipate jurisdictional differences before a product or workflow is scaled across markets.

Where disputes are likely to cluster

The recurring flashpoints are predictable:

• Governance gaps. Problems arise when AI tools are adopted across departments without a clear owner or escalation path.
• Rights and consent. Unclear documented authority to use name, image or likenesses increases risk.
• Trust and accuracy. Hallucinations, misleading summaries or synthetic media can irreparably damage credibility.
• Privacy and personalization. AI systems that rely on personal data or behavioral inferences require clear limits and a process for review, including consent and data minimization where required.
• Labor and creative control. Disputes caused when AI replaces creative work or when talent has not consented to synthetic representations.
• Platform and vendor leverage. Third-party changes can affect traffic, monetization, moderation outcomes and risk allocation.

Conclusion

AI is already part of the media sector. It is changing how companies create, distribute, and monetize content. It is not always inherently dangerous in every use, but it does introduce risks that media companies need to be aware of.

Companies that understand where AI is actually being used, what content and data are being exposed to it, and who has the ability to control or explain the result, will be better positioned to capture value and to respond, when challenged. Mitigating AI risk requires coordination among multiple and often disparate business teams. It also requires vendor contracts and internal processes that reflect how media companies actually operate.

The central issue is not whether AI can create value; media companies are already proving that it can. Instead, media companies should look to whether that value can be captured without weakening their rights and credibility built from years of reputational trust.

Synthetic content has real consequences, but so does operational AI that no audience will ever see. Media companies should not wait for a dispute or inquiry before ascertaining whether their AI use can be defended. In this sector, responsible AI is not just about innovation, it is about preserving and protecting the value of the content and the rights of the people connected to it, without sacrificing the company’s credibility.