What Is Synthetic Media in Advertising?

Synthetic media in advertising refers to any audio, image or video asset that is fully or partly generated or materially altered by artificial intelligence, rather than captured directly from real‑world performance.^[1] This spans relatively familiar tools, such as AI voiceovers and face-swap localisation, through to hyper‑realistic deepfakes. As usage grows, so do expectations around disclosure, provenance and consent, shaped by technical standards such as C2PA content credentials and new regulation including the EU AI Act and emerging UK guidance.^[2][3]

Definition and scope of synthetic media in advertising

In advertising practice, synthetic media is typically defined as any image, audio or video asset that has been generated or materially altered using AI models, including generative adversarial networks, diffusion models and neural text‑to‑speech systems.^[1][4] The defining feature is that key visual or sonic elements are synthesised by an algorithm, not recorded from a live performance. This includes AI‑generated spokesperson avatars, cloned voices, composited crowds or AI‑extended background scenes. Text‑only outputs, such as copy suggestions, are usually treated as a creative input rather than on‑screen or on‑air synthetic media.

The spectrum runs from light‑touch augmentation to fully synthetic characters. At one end sit style filters, AI clean‑up or background replacement that subtly improve or adapt captured footage. Further along are AI‑driven AI casting workflows, where models generate photorealistic faces or bodies that were never filmed, or localise an existing performer’s lip‑sync into new languages. At the far end are complete deepfake performances, where a subject’s likeness, expressions and voice are reconstructed frame by frame without additional live action.^[1][4]

From voice cloning to deepfakes

Voice cloning is one of the most common forms of synthetic media in commercial use. Modern neural text‑to‑speech and voice conversion systems can reproduce a specific speaker’s timbre and prosody from relatively short recordings.^[5] In advertising, this enables scalable localisation, iterative script changes after recording, or synthetic brand voices that speak dozens of languages without repeated studio sessions. Similar techniques generate crowd beds, character performances and dialogue for digital avatars, often with no on‑mic talent present at all.

Deepfakes combine several of these techniques. Face‑swap models and neural rendering systems can map one person’s facial movements onto another’s likeness, or generate a wholly artificial person who appears live‑action and photorealistic.^[4][6] For advertisers, this opens up applications such as virtual influencers, digital doubles for stunts, and AI‑disclosed testimonial formats. The same techniques, however, are associated with impersonation, political misinformation and non‑consensual imagery, which is why many brands and regulators treat deepfake‑style executions as a higher‑risk category requiring stronger controls.^[6]

Content provenance and the C2PA standard

As synthetic media becomes easier to produce, content provenance has emerged as a priority. The Coalition for Content Provenance and Authenticity (C2PA), a cross‑industry standards group backed by organisations such as the BBC, Adobe and Microsoft, has created an open technical specification for attaching “content credentials” to digital media.^[2] These credentials cryptographically record how a given asset was created and edited, including whether AI tools were involved, in a format that can be verified by compatible platforms and tools.

For advertisers and production partners, C2PA offers a way to embed audit trails at the file level, rather than relying solely on production paperwork. Content credentials can record the use of synthetic faces, AI‑generated backplates or cloned voices, along with timestamps and tool identifiers.^[2] While adoption is still emerging, agencies and broadcasters are beginning to explore C2PA as a practical enabler for brand safety, talent‑rights management and regulatory compliance, particularly for campaigns that make heavy use of AI‑generated elements.

Regulatory and ethical considerations

Regulation is starting to address synthetic media directly. Under the EU AI Act, providers and users of certain AI systems must clearly disclose when people are interacting with AI systems, when content such as images or audio is artificially generated or manipulated, and when emotion recognition or biometric categorisation is used.^[3] Article 50 sets out specific transparency obligations for deepfakes, requiring that users “disclose that the content has been artificially generated or manipulated” unless an exemption applies, for example in satire that is appropriately contextualised.^[3]

In the UK, the Advertising Standards Authority (ASA) and Committees of Advertising Practice (CAP) have stated that existing rules on misleadingness, identification of advertising and offence apply regardless of whether media is AI‑generated. The ASA has indicated that ads using AI, including deepfakes or synthetic influencers, must not materially mislead about the identity, endorsement or capabilities of a product, and should make any significant artificiality clear where it affects consumers’ understanding. Ofcom has highlighted similar concerns for broadcasting and online safety, particularly around news‑like content and impersonation.

Industry initiatives are also shaping norms. The Partnership on AI’s “Responsible Practices for Synthetic Media” framework recommends informed consent for any replicated identity, clear disclosure where synthetic media may influence trust, and internal risk assessments for high‑impact uses.^[6] For advertisers, this typically means robust talent contracts for likeness and voice rights, clear internal labelling of AI assets, and appropriate on‑screen or in‑copy disclosures wherever synthetic elements could materially influence how viewers interpret a message.