Hire a Brand Safety and Synthetic Media Lead Before the Deepfake, Not After

Whose Desk Does a Deepfake of Your CEO Land On?

A sixteen-second clip of your chief executive endorsing an investment scheme is live on three platforms by 7am. The clone is good enough that two board members ask whether it is real. The first four people who see it each assume somebody else is handling it, so for ninety minutes nobody is. That gap is the seat, and it does not close by itself.

A brand safety and synthetic media lead is the standing owner of both directions of this problem. Inbound: impersonation of executives, products and support channels, counterfeit ads running the logo, and what an AI answer engine says about the company when nobody from the company is in the room. Outbound: the rules for creative the company generates itself, which means consent for any likeness or cloned voice, disclosure where a viewer would otherwise be misled, and a named human review that no launch date is allowed to compress.

The reason this stopped being an ad-ops setting is volume on one side and cheapness on the other. Industry analysis of 2026 brand safety describes AI-generated content and shifting platform moderation as having moved the topic from a media-buying checkbox to a board-level risk ¹. A 2026 practitioner survey reported that 53 percent of US media experts named ad proximity to generative AI content a top media challenge for the year, against forecasts that a large majority of web content will be machine-generated ². Meanwhile the cost of a convincing fake of a named person fell to the price of a subscription.

What makes the job a hire rather than a policy document is the pre-launch part. The practice that works now is upstream: creative, partnerships and product surfaces get tested for permission before they ship, not measured after ³. That is a review gate with a person behind it, and a gate belongs to somebody or it is a suggestion.

One scoping decision before you write the description. If most of your exposure is the paid-media side, this can start as a senior manager inside growth marketing, adjacent to whoever runs your growth experimentation. If the exposure is impersonation of people and products, the seat needs a comms and legal reporting line from day one, and the growth org is the wrong parent.

Which Tells Separate a Real Synthetic Media Lead From a Policy Writer?

Everyone interviewing for this title can describe a policy. Fewer have made a decision that cost the business something and can tell you the number. The separating question is not what they believe about AI creative. It is what they have refused, what it cost, what they conceded to get the refusal accepted, and what the team did the following quarter.

Six tells that hold up in an hour:

• They ask about consent paperwork before they ask about tools. A cloned voice or a synthetic spokesperson raises a rights question that no detection product answers. Real ones want to see the talent agreement language first.
• They separate disclosure from deception. Not every AI-made asset needs a label, and a candidate who says all of them do has not run a campaign. Ask where the line sits and listen for a rule a copywriter could apply without calling them.
• They have a takedown motion, with timings. Which platform form, what evidence attaches, who signs, how the executive is briefed while it is pending, and what goes out publicly in the first hour. Vagueness here is the loudest signal in the interview.
• They treat the agency and the tool vendor as part of the surface. Most generated creative arrives through somebody else's pipeline. Real candidates ask what the contracts say about training data, indemnity and model provenance, which is the same conversation your vendor risk work already runs.
• They can name a rule they got wrong. A brand safety exclusion list that quietly defunded legitimate news, a keyword block that starved a category, an over-broad hold that pushed a team to route around them.
• They ask who else can override them, and how the override is recorded. Candidates who have held this seat before always ask. It is the question that predicts whether they will still be effective in month nine.

One anti-tell worth naming plainly. A candidate who proposes to solve this by detecting AI-generated content, whether in a competitor's ads or in an applicant's writing sample, is selling a capability that does not hold up. The workable version of the job is provenance, consent, disclosure and response, all of which are records the company controls.

Which Backgrounds Produce This Lead, and How Did They Get Good?

The obvious feeder is media buying, where somebody already owns inclusion lists, adjacency standards and the awkward call to a publisher. It is a fine start and a narrow one, because it trains judgment about where an ad lands and almost none about a fake of a person. The stronger hires usually come from somewhere that already had to decide, in public, on a deadline, whether something was real.

Four backgrounds that transfer well and get overlooked. Platform trust and safety investigators, who have run impersonation and coordinated-inauthenticity queues at volume and know what evidence a takedown actually needs. Newsroom standards editors and verification desks, who have spent years deciding what to publish about a video of uncertain origin, and who bring a habit of writing down the reason. Music, film and talent rights administrators, who already handle likeness and voice consent as ordinary paperwork rather than as a novel crisis. And crisis communications leads who have run a hostile news cycle and can hold a line while the facts are still moving.

How the good ones got good with AI is worth asking directly, because the answer is diagnostic. Ask what they have generated themselves and what it got wrong. The useful answers are concrete: cloning their own voice to find out how few seconds of audio it took, running the company's brand-voice prompt until it produced a claim legal would never approve, generating a batch of ad variants and finding the two that misstated a product capability, or asking a model what it says about their employer and discovering the answer was drawn from a three-year-old press release. That habit is the job in miniature. Someone who has never made this material misjudges what is easy, and someone who trusts what it produces fails more expensively.

The adjacent internal candidate is worth interviewing before you go outside. Somebody in your marketing org has already been absorbing this work unofficially, usually the person the agency emails when a script feels risky. They know which campaigns nearly shipped. Promoting them and backfilling underneath is often faster than an external search, and pairs naturally with whoever owns the context and prompts your teams reuse, since half the outbound rules end up encoded there.

Source Candidates Where Impersonation Already Gets Handled

Post where people have already worked a queue of fakes. Platform trust and safety alumni networks, the Brand Safety Institute's certified practitioner community, IAB and Media Rating Council working groups, the Content Authenticity Initiative and C2PA provenance community, and the standards and verification circles that grew out of newsrooms. Those venues return people who have defended a decision. A generic marketing board returns people who have read about the problem.

Feeder employers follow the same logic. The large platforms and ad exchanges staffed impersonation and ad-integrity teams years ago. Verification and brand safety vendors have analysts who read this material all day. Talent agencies and studios have rights administrators fluent in likeness consent. Consumer brands in categories that attract counterfeits, meaning luxury, supplements, crypto-adjacent finance and consumer electronics, have quietly built the muscle already.

Search on the duty rather than the noun, because the title has not settled. AI brand governance manager, synthetic media risk manager, brand integrity lead, media quality lead, ad integrity manager, trust and safety manager for brand. Each shows up attached to a different parent function.

Then screen on artifacts, not on interest. Ask every candidate for something they wrote that other people had to follow: a creative review standard, a takedown playbook, an inclusion list rationale, an incident write-up with the timeline in it. Read it before the conversation. This is a writing and judgment job, the samples exist, and one real document tells you more than a portfolio of campaigns they were near.

How Do You Close This Hire, and Should the Seat Sit On-Site?

Close on authority first, money second, and say both out loud in the first conversation. The people worth hiring have all watched a safety seat get overruled by a launch date, so name the reporting line, name exactly what they can hold, and describe how an override gets written down and who sees it. An offer that is vague on that line loses to an offer that is specific on it, even at lower pay.

On compensation, be honest about the state of the evidence. No government wage series covers this title, and no credible published band for a brand safety and synthetic media lead was found while writing this, so any single number here would be invented. What can be said without a source is structural: price the seat against the internal band it most resembles, which is usually senior manager to director in marketing, and check it against two adjacent markets that are genuinely competing for the same people, platform trust and safety and crisis communications. Both pay above general marketing for the same seniority, and both will be bidding. If you must anchor, anchor on your own offer history for a seat with equivalent veto authority rather than on a scraped figure.

What kills the offer is predictable. A reporting line through the growth team whose campaigns they are meant to review. A description that turns out to mean writing a policy nobody enforces. No budget for the response side, meaning monitoring, outside counsel and a takedown vendor, which reads accurately as a statement about how serious the mandate is. And a hiring process long enough that a platform recruiter closes them first.

On location, most of the work travels. Standards writing, creative review, provenance and vendor conversations are all fine remote, and postings for adjacent titles lean remote or hybrid. Three parts do not travel well. A live impersonation incident wants the lead, comms and counsel in one room for the first day. Pre-launch review of a major campaign goes faster in person with the creative team than in a comment thread. And the first ninety days are a discovery exercise, finding which teams and agencies are already generating what, which goes badly over video with people who have no reason to trust you yet. Remote with scheduled on-site weeks and a named expectation for incident response is the arrangement to write into the offer.

See the benchmarks >