When a manager first floats the idea of using AI video for training or internal communication, the pushback is almost always framed the same way: "people won't trust it, it feels fake." That's the objection on the surface. It's rarely the objection underneath.
After enough conversations with companies rolling out AI video internally, a pattern shows up. The resistance usually breaks into three separate issues, and they need three separate answers.
1. Ownership and image rights. If an employee's likeness or voice is used to build an avatar, who owns that digital version? What happens to it if the employee leaves the company? This is a real legal question, not a feeling, and it deserves a real contract answer, not a reassurance.
2. Job displacement fear. Employees in training, communications, or video production roles sometimes read an AI video initiative as a signal that their job is being phased out. This fear doesn't go away by explaining that the technology is impressive. It goes away when leadership is specific about what the AI is and isn't replacing.
3. Format trust, also known as the uncanny valley. This is the one people usually mean when they say "it feels fake." An avatar that almost looks human but doesn't quite move right creates a small, persistent unease that undermines the message no matter how good the script is.
Treating all three as one problem, "people don't trust AI," leads to generic reassurances that don't land, because you're answering the wrong question for two-thirds of the room.
A common instinct is to make the AI video as seamless and human-passing as possible, on the theory that if people can't tell it's AI, they won't object to it. This tends to backfire. When employees later find out a video was AI-generated and weren't told, the trust damage is worse than if the AI had been disclosed from the start, even if the video itself was well made.
The more durable approach is the opposite: disclose AI use proactively, every time, not just when asked. Say plainly that this is an AI-assisted video, and say who directed it, wrote it, and signed off on it. Counterintuitively, this tends to reduce the trust problem rather than expose it. People aren't objecting to AI. They're objecting to being misled about it.
The most effective reframe isn't a defense of the technology. It's a clear description of what the AI is actually doing. AI can generate a voice, animate an avatar, translate a script into another language, or produce b-roll. What it does not do, in a well-run production, is decide what the message should be, how it should land, or whether it's ready to publish. That's a human director's job, start to finish.
Making this distinction explicit, rather than assuming people already understand it, does more to resolve the authenticity objection than any amount of technical explanation about how realistic the avatar is.
Companies that try to roll out AI video company-wide in one move tend to face more resistance than companies that start with a single, low-stakes pilot. A benefits explainer or an IT walkthrough is a reasonable place to test the format, gather employee reactions, and adjust before using it for something higher-stakes like leadership communication or compliance training.
A pilot also gives HR and comms teams something they rarely have in these conversations: actual internal data on how their own employees responded, rather than a general argument about AI video trends.
The companies that get the most value out of AI video aren't the ones that convince employees the technology is flawless. They're the ones that are honest about what it is and clear about who's still in charge of it.
