Industry Trends

By Jed Williams, Co-founder, Spexster

Camera operator filming a real product on set while an editor's screen beside them shows a transcript timeline, illustrating AI in video production — Editorial illustration of AI in video production: a camera operator films a real product while an editor works from a transcript timeline.

AI in Video Production: What It Changes for Brands Hiring Crews

AI now speeds up the work around a shoot (scripts, storyboards, transcript editing, captions, translation, the occasional generated B-roll shot) but does not replace filming your real people, locations and product. The practical changes for brands are in the brief and the contract: where AI may be used, written consent for any voice or likeness work, and knowing that purely AI-generated material may not be protected by copyright.

The pitch says AI will do half the shoot. Which half?

Every brand team has now sat through a pitch where a generated clip of a mountain road or a glowing product render was presented as the future of their video budget. The clip looked good. The question nobody in the room could answer was: how much of our actual film, the one with our CEO, our factory and our product in a customer's hands, can this make?

The honest answer in 2026 is: very little of what goes on screen, and quite a lot of the work around it. AI is now in the script draft, the storyboard, the transcript edit and the captions, and that is a real change in how fast a cut comes back. It is not in your product shot, your interview or your event, and a growing body of law says that where a real person's face or voice is involved, it cannot be without their signed say-so. This guide maps where the line sits, task by task, and what to write into the brief and the contract so you are on the right side of it.

Where AI actually sits in a production in 2026

The tools move fast, and some have already come and gone:

Google Veo 3.1. Paid preview since October 2025 via the Gemini API, Vertex AI, the Gemini app and Flow. Generates native audio, takes up to three reference images and can extend scenes to a minute or more.
Runway Gen-4.5. Released December 1, 2025. Runway itself notes the model can show effects before causes and that objects can vanish or appear between frames.
OpenAI Sora. OpenAI shut the Sora app on April 26, 2026 and ended the API on September 24, 2026. If a pitch still leans on Sora, ask what replaced it.
Adobe Firefly video. Adobe says its own Firefly Video Model is trained on licensed and public-domain content and is "designed to be safe for commercial use," while partner models inside Firefly, such as Veo 3.1 and Runway Gen-4.5, "may have different usage terms." Most models generate clips of up to 8 seconds, some up to 15.

That is the real shape of generative video for brand work: short clips for an establishing shot, a background plate, a transition or a concept test. Not a two-minute brand film with your CEO, your factory floor and your product in a customer's hands.

AI in pre-production: faster drafts, same decisions

Pre-production is the lowest-risk place for AI: nothing generated here ends up on screen.

Scripts and shot lists. Language models draft a script, interview questions or a shot list from a brief. A director still has to make it shootable: knowing a line won't work in a noisy warehouse, or that a three-location day is really two.
Storyboards. Adobe's Firefly Boards turns script descriptions into storyboard frames with consistent lighting and character style, so a brand and a director can agree on a look before anyone books a gaffer.
Previsualization. A generated clip can show how a camera move might feel. Treat it as a sketch, not a promise.

The hard calls stay the same: what the story is, who is on camera and where you can film.

AI in post-production: the biggest practical change

Transcript editing. Premiere's Text-Based Editing transcribes footage automatically and lets an editor cut an interview by editing the text, including finding and deleting pauses. DaVinci Resolve 20's IntelliScript goes further: give it a script and it matches transcribed audio to the lines and builds a timeline, with alternate takes on separate tracks.
Captions and translation. Adobe says Premiere's caption translation covers 27 languages; Resolve 20 generates animated subtitles. For multi-market edits, that removes a slow step.
Searching footage. Premiere's Media Intelligence recognizes objects, locations and camera angles across large libraries.
Fixing gaps. Premiere's Generative Extend adds frames and audio to the head or tail of a clip, in 4K and vertical. Resolve 20 adds speaker-based multicam switching and an audio mix assistant.

A good editor using these tools turns cuts around faster, but pace, story and what to leave out are still their judgment. When you brief a freelance video editor, ask which of these tools they use and whether any generated frames or audio end up in your final masters.

Where AI helps and where you still need a crew

| Task | Where AI helps | Where you still need a crew | |---|---|---| | Script and shot list | First drafts, interview questions, rewrites for length | Making it shootable on your locations, schedule and budget | | Storyboards and look | Fast frames and mood boards for sign-off | A director of photography who can light and lens it for real | | Interviews and testimonials | Transcription, transcript editing, captions | Real people on camera, clean sound, lighting skin tones well | | Product video | Background plates, abstract transitions | Accurate product shape, color, texture and scale; your actual product | | Locations and B-roll | Short establishing or concept shots | Your real store, office, factory or city, and the permits to film there | | Events | Fast searching and assembly of footage | Coverage of something that only happens once | | Edit and versioning | Rough assembly, captions, translation, extending clips | Story, pace, brand judgment and final sign-off |

AI helps most with work that can be described in words or repaired after the fact. It cannot film something true about your brand, and audiences watching a testimonial or product demo trust that what they see happened.

Union talent. The 2025 SAG-AFTRA Commercials Contracts require producers to get "clear and conspicuous" consent to create or use a performer's digital replica, given separately from the employment contract, with a "reasonably specific description" of the intended use and advance notice. Without further written consent, replicas have to be deleted after the commercial's usage period.

State law. California's AB 2602 makes a contract provision for a digital replica of a performer unenforceable for performances fixed on or after January 1, 2025, unless it describes the intended uses reasonably specifically and the performer had legal representation. California's AB 1836 covers digital replicas of deceased personalities. Tennessee's ELVIS Act, signed March 21, 2024, protects voice as well as likeness against unauthorized AI cloning. New York's synthetic performer law, in effect since June 9, 2026, requires advertisers to conspicuously disclose AI-generated synthetic performers in ads, with civil penalties of $1,000 for a first violation and $5,000 for later ones.

Platform labeling. YouTube requires disclosure of realistic AI content that makes a real person appear to say or do something they didn't, alters footage of a real event or place, or shows a realistic scene that didn't happen; scripts, captions and color correction are exempt. Meta labels ads significantly edited with its own generative tools and says it detects third-party AI tools through industry-standard signals, showing an "AI info" label.

Practically: everyone on camera should sign a release saying whether their voice or likeness may be used with AI, and for what. Our guide to talent and location release forms covers the basics.

Performer on set with a release form on a clipboard in the foreground, representing consent for voice and likeness use — Editorial illustration of performer consent for AI voice and likeness use, with a signed release form on set.

Who owns AI-generated footage?

The US Copyright Office's January 2025 report on copyrightability concluded that prompts alone are not enough for authorship. AI-assisted work can be protected where a human makes creative arrangements or modifications, but material whose expressive elements are determined by a machine is not. Footage your crew shot and edited is protectable as normal; a fully generated clip in the edit may not be yours to stop others copying. If ownership matters, keep generated material in supporting roles and know where it was used.

How to brief crews about AI use

Add a short AI section to every brief:

Allowed. For example: transcription, captions, translation, footage search, rough assembly.
Needs approval. Generated shots, Generative Extend on key footage, AI voice work that changes what someone said.
Not allowed. Generating or altering your product, people or locations without sign-off; uploading unreleased footage to tools whose terms allow training on it.
Disclosure. A list of any generated or AI-altered material in the delivery, so you can label it where platforms or state law require.

What to put in the contract

On top of the usual deliverables, dates and rights, add:

An AI-use clause mirroring the brief's allowed, approval-needed and prohibited uses.
A disclosure obligation to identify generated or AI-altered material in delivered files.
Data handling: which tools your footage may be uploaded to.
Releases for everyone on camera, with specific digital replica and voice language.
Ownership terms that account for AI-generated elements.

For other shifts this year, see video production trends.

How Spexster fits

Whatever you decide about AI, it belongs in the brief, and the crew needs to have agreed to it before the shoot, not after a generated frame turns up in the master. On Spexster that brief goes to vetted videographers, cinematographers and video editors, each reviewed before appearing with no paid placement, who apply to the job having read your terms. With White Glove, a dedicated project manager writes your AI-use rules into the contract alongside usage rights and revision rounds, then handles sourcing, scheduling, logistics and escrow payment through Stripe at no additional fee. The part AI cannot do, putting a skilled person in front of your real product in your real location, is the part Spexster was built for. Read how Spexster has helped brands and agencies in our case studies.

Frequently asked questions

Can AI replace a video crew for brand content?

Not for most brand work. Generative tools make short clips suited to establishing shots and concept tests, but cannot film your actual people, product or locations. AI mainly speeds up planning and editing.

Do I need to disclose AI-generated content in my video ads?

Often, yes. YouTube requires disclosure of realistic AI content involving real people, real events or realistic invented scenes, but not script or caption help. Meta labels ads made with generative AI, and New York requires conspicuous disclosure of AI synthetic performers in ads.

Can I copyright a video made with AI?

The human-made parts, yes. The US Copyright Office says prompts alone do not make you an author, so purely AI-generated material may not be protected, while crew-shot footage edited with AI tools is.

What should a contract with a video crew say about AI?

Allowed, approval-needed and prohibited AI uses; a duty to flag generated or AI-altered material; rules on which tools your footage can go into; and releases with specific digital replica terms for voice and likeness.

Is Adobe Firefly video safe for commercial use?

Adobe says its own Firefly Video Model is trained on licensed and public-domain content and designed to be commercially safe. Partner models inside Firefly may have different terms, so check which model made a clip before it goes in an ad.