Hollywood Assistants Are Using AI Despite Their Better Judgment

The Shadow Workflow in Hollywood

A low-level anxiety is spreading through the administrative wings of Los Angeles. The source is not the writers' rooms or the production lots, but rather the desks of the assistants who keep the machine running. Faced with expanding workloads and shrinking headcounts, a significant portion of Hollywood support staff has begun integrating artificial intelligence into their daily routines. They are proceeding despite better judgment, often without company approval, and frequently without security oversight.

The situation does not fit the narrative of AI replacing jobs overnight. Rather, it describes the friction of modern work forcing employees to bypass protocol just to get through the day. When output demands exceed headcount capacity, the path of least resistance wins. In Hollywood, that path increasingly leads to an open chat window with a large language model.

Efficiency is only part of the risk; the real danger lies in intellectual property. Industry observers note the concern centers on personal AI tools rather than enterprise accounts. When an assistant pastes a script treatment, a character breakdown, or a production schedule into a public model, that data becomes part of the model's training set. For a studio, this creates a significant liability concern where proprietary concepts leak into the public domain. For the assistant, it represents a career gamble. They are effectively feeding the creative work they are paid to protect into an ungoverned system to save themselves an hour of drafting.

The Fear of Teaching Your Replacement

Industry reports frequently overlook the psychological toll of this shadow workflow. Assistants are not just using tools; they are navigating a minefield of expectations. The pressure to adopt AI comes from executives who want the output but not the risk.

One studio assistant captured the tension in a candid moment: "When they say, 'You should be using AI,' the first thought in your head is: 'Are you asking me to teach you how to replace me with technology?'"

This hesitation is rational. If you can automate the scheduling, the drafting, and the research that makes your job unique, what is left for you to do? Yet, the work must get done. Industry watchers fear the encroachment of automation, but they also acknowledge that everyone in Hollywood is using AI, even if they are scared to admit it. This silence creates a culture where mistakes go unreported. If an assistant makes a factual error generated by a model, they are unlikely to flag it to leadership because admitting they used the tool might get them fired.

Quality control degrades as a result. Scripts and schedules are being polished by algorithms that hallucinate facts, and the humans reviewing them are too afraid to disclose the source of the error. This creates a feedback loop where bad data enters the system, gets validated by human oversight, and then circulates as truth. The studio accepts the output without knowing the provenance, meaning errors become embedded in the production timeline before anyone realizes the tool was used.

The Organizational Gap

Management, rather than the technology itself, represents the core failure. There is little to no training offered to assistants on how to use AI safely. Companies are either banning it outright, which forces it underground, or allowing it without guardrails, which invites disaster. Neither approach solves the workload problem.

When support staff lack a structured way to manage information, they turn to the easiest tool available. A public chatbot requires no setup, no login, and no configuration. It is frictionless. Friction often serves as a necessary safety feature. When there is zero friction to dumping a confidential document into a web browser, confidentiality evaporates.

This is where the distinction between a chaotic workflow and a managed one becomes critical. The solution isn't necessarily to ban the tools, but to create an environment where data is organized before it ever meets an AI. You cannot secure what you do not track. Studios need to define exactly what data can be processed externally and what must remain internal. Without clear boundaries, employees will default to the path that offers the quickest resolution to their immediate problem, regardless of the long-term security implications.

Structuring the Chaos with Notion

For professionals trying to navigate this gray area, the difference between a liability and a workflow often comes down to where the data lives. Tools like Notion offer a way to manage the information intake without immediately surrendering it to a generative model.

Notion is primarily a workspace and database tool, but it is becoming the standard for the infrastructure that supports AI use. By using a workspace like Notion, an assistant can draft, organize, and store scripts or schedules within a secure, private environment before any external tool touches the data. It allows for version control and clear ownership, which is missing when you just paste text into a chat window. This creates an audit trail that proves who handled the data and when.

There is a significant trade-off involved. Setting up a robust Notion workspace requires time. You have to build the database, configure the permissions, and train yourself on the system. It is not frictionless. For an assistant drowning in emails and urgent calls, the time required to organize their workflow might feel like a luxury they cannot afford. This is the reality of the trade-off: security costs time.

Companies wishing to prevent the "ungoverned LLM" scenario must provide the infrastructure that makes security faster than the insecure option. If the secure path takes twenty minutes and the insecure path takes two, the insecure path will win. Leadership must understand that providing the right tools is not just an IT expense; it is a risk mitigation strategy that requires upfront investment in setup and training.

The Trade-off of Automation

Even with a structured system, the introduction of AI into creative workflows creates specific downsides that are rarely discussed in the press. Financial metrics do not capture the primary cost, which is the erosion of institutional memory. When an assistant relies on an AI to summarize a script or pull notes from a meeting, they are offloading the cognitive labor required to understand the material.

Over time, the assistant becomes a conduit for the AI's output rather than a guardian of the project's details. They stop knowing the material intimately because they know where the AI put it. If the tool fails, or if the data is lost, the human element is too weak to catch the fall. This lack of deep familiarity means that when a producer asks a nuanced question about a character's arc or a scheduling conflict, the assistant cannot answer from memory but must query the system again, slowing down decision-making.

The learning curve for these tools remains steep. Notion, for example, is powerful but can be overwhelming. It requires a shift in thinking from linear document management to database management. For a support staff already stretched thin, adding a new layer of software complexity can lead to burnout. The tool is meant to save time, but the implementation often steals it first.

There is also the issue of dependency. Once a workflow is built around specific AI integrations, removing them becomes difficult. If the company decides to switch vendors or shut down the enterprise account, the assistant's entire productivity system collapses. This locks the worker into a specific tech stack they may not have chosen, reducing their portability in the job market. Skills learned on a proprietary platform may not translate to other studios, limiting their career mobility.

The Bottom Line

Hollywood assistants are using AI because the workload demands it, not because the technology is ready. The industry is facing a security crisis driven by a management failure. By failing to provide approved, secure tools and training, studios have pushed their most vulnerable employees into the shadows of "shadow IT."

Moving forward requires acknowledging that AI is not a magic wand for workload management. It is a risk multiplier that requires strict governance. Tools like Notion can provide the necessary guardrails, but only if companies invest the time to set them up properly. Until the secure option is as fast as the insecure one, assistants will continue to feed scripts into public models to get their work done. The question is no longer if AI is in the room, but how much of the creative process is now owned by an algorithm that no one at the studio is accountable for.

Sources: https://www.hollywoodreporter.com/movies/movie-features/hollywood-assistants-ai-development-1236553905/ | https://www.yahoo.com/entertainment/movies/articles/hollywood-assistants-using-ai-despite-200000291.html | https://www.reddit.com/r/boxoffice/comments/1sd9k4v/hollywood_assistants_are_using_ai_despite_their/ | https://www.hollywoodreporter.com/movies/movie-news/hollywood-ai-artificial-intelligence-cannes-1235900202/