It happens more often than you'd think. You involve someone from the company — a salesperson, a technician, the founder — to record a webinar, shoot a product video, take photos for the campaign. The content works, spreads, brings leads. Then that person leaves the company. And at that point comes the awkward question: can we still use that material with their face?
It's exactly the situation we found ourselves discussing recently: a webinar recorded with a salesperson who has since changed companies. In that case we're talking about a single video. But the problem becomes much more serious when a company has built months of content around a person — dozens of videos, a webinar series, an entire YouTube library, testimonials on the site — and that person leaves.
The short answer is: it depends, and almost always less than you'd want. The long answer deserves this article, because this isn't just professional etiquette: there are rights that Italian law protects fairly rigorously, and handling them poorly can cost in money, time, and reputation.
Two different rights that are often confused
The first step is to understand that in a video with an employee two distinct rights coexist, and it's precisely the confusion between the two that generates most problems.
On one side there's the ownership of the content: who owns the file, the edit, the recording. If the webinar was shot by the company or commissioned by it, normally the company holds the exploitation rights. On this, usually, there's no dispute: the video is theirs.
On the other side there's the image rights of the person filmed, and it's a whole other story. In Italy this right is protected by article 10 of the Civil Code and by articles 96 and 97 of the Copyright Law (L. 633/1941). The basic rule is simple: a person's portrait cannot be displayed, reproduced, or put on the market without their consent, except for narrow exceptions (notoriety of the subject, judicial purposes, scientific or educational purposes, public events). Commercial and marketing use doesn't fall under any of these exceptions.
Translated: owning the video doesn't mean having the right to show the face of whoever's in it. They are two separate things. You can be 100% owner of the file and still not be able to use it, if consent from the person filmed is missing or has been withdrawn.
The point that catches almost everyone off guard: consent is always revocable
Here comes the aspect that, in our experience, surprises most business owners. People tend to think: "We signed a contract, they agreed, so we're all set." That's not the case.
Italian jurisprudence is consolidated on a precise principle: consent to the exploitation of one's image is an autonomous and always revocable act, distinct from the contract that contains it. The Supreme Court affirmed it in multiple rulings (judgments no. 1748/2016, no. 27506/2008, no. 3014/2004) and the Court of Appeals of Milan reiterated it with a ruling of March 23, 2021.
The judges' reasoning is this: one thing is the conditions and modalities of image exploitation, regulated by the contract; another is consent to being filmed and shown, which remains personal and revocable at any time, regardless of the agreed term and regardless of the compensation paid. If the person revokes, the company must cease use. It can eventually ask for damages for contractual breach, but it can't force the person to "stay" in the content.
To this is added the GDPR, because the image of a recognizable person is personal data. Also on the privacy front, consent is revocable at any time: the withdrawal applies going forward (it doesn't make what you did before unlawful, but it blocks use from that moment on). The Data Protection Authority is clear that, at the end of the employment relationship, an employee's personal data — photos included — must be removed from company channels if there's no valid legal basis to support their use.
So, going back to the webinar of the salesperson who left: the fact that they participated willingly when they were an employee doesn't guarantee you can use it forever. If they ask you to take it down, in the vast majority of cases you have to take it down.
Two Italian cases that help understand what's at stake
To make the discussion concrete, two real examples.
The Rocchetta–Insinna case. It's the textbook testimonial case. The mineral water brand Rocchetta (Cogedi/Co.Ge.Di. group) had Flavio Insinna as its public face, with an important contract — around €275,000 a year. In 2017, after some embarrassing behind-the-scenes footage was broadcast by Striscia la Notizia, the company wanted to end the relationship and started a multi-million damages claim for reputational damage suffered (figures between €2 and €2.3 million, according to the press). Insinna, for his part, opposed it saying he was the one damaged.
What does it teach us, even though it involves a famous personality and not an employee? That tying communication to a person creates a mutual and risky dependency: when the relationship goes sour, the content built around that face suddenly becomes a problem, not an asset. And that without well-written clauses you end up in arbitration or in court fighting over who owes what to whom.
The Milan Court of Appeals principle (2021). The case cited in the jurisprudence involves precisely an initially unknown person — a model — whose image had been used by contract. The judge confirmed they could withdraw consent and block use, despite the contract. It's the most insidious scenario for a company: the "any person" filmed in a video today can change their mind tomorrow, and the law gives them the right to.
The common thread in both cases is the same: whoever puts their face on it retains a power over that face that the contract, alone, doesn't turn off.
What you can (and cannot) do with materials already produced
Suppose the employee has already left and you have an archive full of content with them. How do you move?
The first step is to distinguish case by case based on use. One thing is purely informational or historical content (for example a technical webinar where the person explains a product, left in the archive); another is markedly advertising and promotional use, where that face becomes the testimonial of your brand. The second is much more exposed: it's precisely commercial use where protection is strongest.
The second step is to check what you signed. Is there a release? What exactly does it provide — which channels, for how long, for what purposes? Often you discover there's nothing in writing, or that the release was generic and designed for something else entirely. In the absence of a clear agreement, the company's position is weak.
The third step, when the relationship ended well, is the simplest and most underrated: ask. An email in which the former collaborator confirms in writing that they consent to the continued use of specific content, for defined purposes and duration, solves much of the problem. People often say yes without difficulty — but you need to have the confirmation in black and white.
If instead the person opposes, or the relationship closed on conflictual terms, the wiser path is to withdraw or replace the content. Yes, it's annoying to throw away an asset that was working. But the risk of a dispute — takedown request, cease-and-desist, possible damages — isn't worth it.
Best practices: prevent instead of chasing
Everything above is damage management. The real game is played earlier, when the content hasn't been shot yet. Here's how organized companies reduce the problem to almost zero.
Written release, always and specific. Before every production involving a recognizable person, have them sign a release that clearly states: for what content, on which channels (website, YouTube, LinkedIn, ads, fairs…), for what purposes (informational and promotional), and — crucial point — for how long, even after the end of the employment relationship. A generic release is worth little; a specific and well-written one is your first line of defense. Keep in mind, though, the limit mentioned above: not even the best release makes consent absolutely irrevocable. It serves to define the rules of the game and give you a basis for damages, not to "chain" the person.
Distinguish work ownership from image rights. Make sure the contract clarifies that the production (edit, recording, materials) is owned by the company. It's a different point from image consent, but equally important to avoid also arguing over who owns the file.
Map and catalog your content. Keep an inventory of who appears in which materials, where they're published, and which release covers them. It sounds like bureaucracy, but the day someone leaves — or asks you for removal — knowing in ten minutes where to intervene makes the difference between a handled situation and a treasure hunt in panic.
Include content in the offboarding process. When someone leaves the company, alongside credential deactivation and laptop return there should be an item: "content with their image." Clarifying at the moment of exit what stays online, with the person's written consent, avoids surprises months later.
Balance "face-led" content with "evergreen" and brand content. This is pure marketing strategy, not legal. If your entire library rests on the face of a single person, you've built a fragility point. Alternate person-linked formats with content that lives on brand voice, graphics, voice-over, product. That way, if someone leaves, you don't lose half of your content heritage in one shot.
For key roles, consider "buyout" clauses and extended duration. When you invest significant amounts on content built around a person (the founder, a historic face, a star salesperson), it's worth explicitly and generously defining the post-relationship usage duration and, where it makes sense, a compensation that remunerates that prolonged use. It doesn't eliminate revocability, but it drastically reduces the probability that someone has interest in revoking.
The mistakes not to make
To close, the most common slip-ups we see repeated.
The first is to take consent for granted: "they were happy to do it, so we can use it forever." False, as we saw. Consent is revocable and must be documented, not assumed.
The second is to confuse "we own the video" with "we can use it". They are two different rights. Having the file isn't enough.
The third is to have nothing in writing, or have generic releases collected once and never updated. When you need them, you'll discover they don't cover the case in front of you.
The fourth is to keep everything online out of inertia after a conflictual exit, hoping no one notices. If the person left angry, that's precisely where the risk of a cease-and-desist is highest — and a forced takedown, perhaps with legal aftermath, costs much more than handling the matter in time would have cost.
The fifth, more strategic, is to put all eggs in one person's basket. Even legally impeccable, a content library entirely built on one face is a business risk: that person, sooner or later, will leave.
And then AI arrived: regenerate instead of remove
Up to now the reasoning is the classic one: if a piece of content can no longer be used, you take it down or reshoot it from scratch. But in the last two years a third path has opened that changes the terms of the problem. Today, with artificial intelligence tools, content doesn't necessarily have to be thrown away: in many cases it can be regenerated.
What makes this possible, in practice. You can replace the face of someone who's no longer there with that of another person or a synthetic avatar; you can redub a webinar with an AI voice keeping the script, slides, and editing intact; you can rebuild from scratch the figure of the presenter. In other words, the value of the content — the knowledge, the structure, the editing work — is saved, and you only change the face or voice that became a problem. For a library built on one person alone, it's a way out that didn't exist until yesterday.
Careful, though: AI doesn't get around consent, it shifts it. Here the distinction is crucial, and it flips intuition. Using AI to remove the person — replacing their face, replacing their voice — solves the image rights problem, because that face simply isn't there anymore. Using AI to recreate the person — cloning their face or voice to have them say new things — creates a bigger problem than the starting one.
Cloning the image or voice of a recognizable person is to all effects a use of their portrait, and it requires their consent exactly like the original recording. Doing it without authorization isn't a shortcut: it's an unauthorized synthetic portrait, potentially a deepfake, with legal and reputational exposure far greater than that of an old video left online.
A new layer of rules. Beyond image rights and GDPR — which also apply to synthetic content, because a person's face and voice remain personal data — the European AI Act has entered play, imposing transparency obligations on content generated or manipulated with artificial intelligence: deepfakes and synthetic media must be declared as such. Regenerating content, then, isn't just a technical matter: it brings new obligations.
The practical rule. AI is an excellent tool for subtracting a person from a piece of content and keeping its value alive; it's a minefield when you use it to simulate them. And there's a direct consequence on tomorrow's releases: alongside channels, purposes, and duration, it's now worth including an explicit clause on the use of voice and image for generative AI purposes — whether that face may be cloned, redubbed, or regenerated, and under what conditions. It's the updated version of the same care as always: putting in writing beforehand what would otherwise become a dispute afterward.
In summary
The webinar with the face of the salesperson who changed companies isn't an isolated case: it's the small version of a problem that every company doing content marketing will encounter. Italian law strongly protects image rights and makes consent always revocable, regardless of the contract and compensation. Owning the content isn't equivalent to being able to use it.
The good news is that it's an almost entirely preventable risk: written and specific releases, a clear distinction between work ownership and image rights, a content inventory, integration into the offboarding process, and an editorial strategy that doesn't depend on one face alone. Setting it up beforehand costs little. Chasing it afterward costs a lot. And when the damage is already done, AI today offers a way to regenerate content instead of throwing it away — provided it serves to remove the person, not to clone them without their consent.
This article has informational and educational purposes and does not constitute legal advice. For specific cases — especially high-value content or conflictual situations — we recommend the opinion of a professional in copyright and data protection law.
