Amazon MGM Tests AI Tools for Film and TV Production

Amazon MGM Studios is set to begin closed beta testing of its proprietary AI production tools in March 2026, with initial results expected by May. The move marks a concrete escalation in Hollywood's use of AI across the full production pipeline, from early pre-visualization to post-production VFX work.

What the Tools Do

The platform, built under an internal unit called AI Studio, focuses on three primary areas: maintaining character consistency across shots, accelerating pre-production workflows, and supporting visual effects pipelines. The system runs on Amazon Web Services infrastructure and draws on external large language model providers to strengthen its output.

Albert Cheng, Chief Digital Officer at Amazon MGM Studios, has led AI Studio since its creation in the summer of 2025. The unit operates with a mandate to protect the integrity of Amazon's intellectual property while giving production teams faster access to AI assisted tools across the content development cycle.

House of David: The First Large Scale Test

Amazon's biblical epic House of David has already served as the most concrete proof of concept for the studio's AI ambitions. Season 2 of the series incorporated roughly 350 AI generated shots, relying on these tools for scenes that would have required either extensive traditional VFX work or logistically difficult on-set resources.

The Creative Team Shaping Deployment

Amazon has enlisted a group of established industry figures to help refine how the tools are used on real productions. Robert Stromberg, the production designer and director behind Maleficent, brings experience at the intersection of physical and digital filmmaking. Kunal Nayyar, known for The Big Bang Theory, and Colin Brady, a former Pixar animator, round out a team whose backgrounds span performance, animation, and large scale studio production.

Hollywood's Broader AI Shift

Amazon is not alone in this direction. Netflix used generative AI to complete a building collapse sequence in The Eternaut, rendering the complex visual effects work ten times faster than conventional methods. Netflix co-CEO Ted Sarandos has framed the technology as a tool to boost creativity, not just reduce production costs.

The convergence of major studios testing AI at the production level, within the same 12 month window, suggests this is moving from isolated experiment to standard practice faster than many anticipated. Independent filmmakers and smaller productions can access comparable text-to-video and image-to-video capabilities through AI FILMS Studio, without the overhead of an in-house AI unit.

Jobs, Labor, and the Stakes

Amazon's official position is that these tools are designed to assist creative teams, not replace them. That framing sits alongside a pattern that has drawn scrutiny. The company has repeatedly cited AI progress when announcing staff reductions, raising questions about where the line between augmentation and displacement sits in practice.

The issue extends across the industry. Guild negotiations have increasingly focused on how AI tools are credited, how likeness rights are protected, and whether residual structures established for human labor apply when AI takes on portions of the work. The beta results Amazon expects by May 2026 will be watched closely by SAG-AFTRA and the WGA, both of which secured AI provisions in their most recent contract cycles.

Not every major filmmaker is moving in the same direction. Steven Spielberg stated at SXSW 2026 that he has never used AI in any of his films and opposes AI that replaces creative individuals. Amazon's commercial relationship with OpenAI later produced a different kind of constraint: in June 2026, the studio dropped Luca Guadagnino's Artificial, a nearly finished film about the OpenAI leadership crisis, after committing $50 billion to the company in February.

For a broader look at the investment wave driving these developments, see The Hollywood AI Studio Boom: A Breaking Point for Film.

The GenAI Creators' Fund Announcement

At AI On The Lot in May 2026, Amazon MGM publicly committed to a new funding initiative for AI integrated animated series. The GenAI Creators' Fund greenlighted three animated Prime Video series and named several directors attached to the projects. The announcement positioned Amazon as an active investor in AI first content rather than simply a studio testing AI as a production efficiency tool.

The public rollout lasted less than 48 hours before Jorge Gutierrez, director of The Book of Life and one of the named creators, announced his withdrawal. Gutierrez cited concerns about the terms under which AI tools would be integrated into the projects and his understanding of what the program required creatively. His departure illustrated the gap between how studios are framing AI creator programs and how established directors are evaluating the actual conditions.

Albert Cheng and How AI Studio Is Structured

Albert Cheng joined Amazon as Chief Digital Officer before the AI Studio unit was created. His mandate spans digital distribution, platform technology, and the content tools Amazon builds internally. AI Studio operates as a distinct product and engineering function within his organization, responsible for the specific AI tools the studio is testing on productions.

The unit's stated mission, protecting Amazon's intellectual property while improving production efficiency, reflects a specific corporate priority. Amazon's content library includes thousands of hours of original series, major franchise properties, and licensed titles. AI tools trained on third-party content without proper rights clearances create legal exposure, and Amazon's investment in its own tooling is partly a hedge against that risk.

Guild Compliance Requirements on Covered Productions

The 2026 WGA and SAG-AFTRA agreements establish specific requirements for productions using AI on covered work. Writers cannot be required to use AI tools to generate material, and AI generated scripts cannot be credited as written work. SAG-AFTRA's agreement requires performer consent before AI tools are used to alter or replicate a performance.

Amazon's closed beta involves real productions with union crews. How the company is navigating those contractual requirements on active beta productions is not detailed in its public statements. The May 2026 beta results will carry more weight if they include documentation of how guild compliance was handled alongside the technical performance data.

The Guadagnino Withdrawal and What It Reveals

Amazon's $50 billion commitment to OpenAI, announced in February 2026, was the largest single investment by a media company in an AI company to that point. The subsequent withdrawal of Artificial, Luca Guadagnino's film about the OpenAI leadership crisis, connected the financial relationship to a specific editorial decision. A nearly finished film that covered events directly involving Amazon's largest AI partner was pulled from release.

The sequence does not have a simple explanation, and Amazon has not provided a detailed public account of the decision. What it demonstrates is that Amazon's institutional AI commitments and its creative AI program do not operate independently of each other. The company making decisions about what tools its productions use is also the company making decisions about what stories those productions can tell.

What the Beta Results Will Show

The closed beta is measuring whether the character consistency tools perform reliably enough across multiple shots for productions to depend on them, whether the pre-production workflow tools reduce schedule and cost in measurable ways, and whether the VFX pipeline integration works with the software tools production teams already use.

Results expected by May 2026 will not be public. Amazon will use them internally to decide whether to formalize these tools as standard offerings across its production slate. What Amazon standardizes, the broader industry eventually encounters. Independent filmmakers and smaller studios that cannot build internal AI units will adopt whatever tools the major studios normalize as production practice.

The AWS Infrastructure Underneath AI Studio

AI Studio runs on Amazon Web Services, which gives it direct access to the GPU compute infrastructure that generative AI tools require at production scale. That infrastructure advantage is one reason Amazon can build and test AI production tools at a pace that smaller studios cannot match. The compute costs that would be prohibitive for an independent production are internal cost transfers for Amazon.

The AWS relationship also means AI Studio's tools are potentially accessible to third-party productions that license AWS infrastructure. A production company without an internal AI unit could in principle access Amazon's character consistency and VFX tools through a cloud services arrangement, rather than needing to build or license the tools independently. Amazon has not publicly announced that model, but the infrastructure is already aligned for it.

Netflix's Parallel Track and What It Shows

Netflix's use of AI on The Eternaut was the most publicly documented example of major studio AI integration before Amazon's beta announcement. The building collapse sequence, rendered ten times faster than conventional VFX methods, gave Netflix a concrete ROI metric for AI in post-production. Netflix co-CEO Ted Sarandos used that result in investor communications as evidence that AI tools reduce production costs without degrading quality.

Amazon's beta is pursuing similar metrics in different production phases. The focus is on character consistency and pre-production rather than post-production VFX. The two studios are measuring AI's impact on different points in the pipeline, which means the combined results from Netflix's published outcomes and Amazon's eventual internal data will cover a broader range of production use cases than either could demonstrate alone.

The Character Consistency Problem and Why It Matters

Character consistency across shots is one of the clearest practical pain points in AI assisted visual storytelling. Current generative video models produce high quality individual outputs but struggle to maintain a character's appearance, clothing, and distinguishing features across multiple shots of the same scene. A character who looks slightly different between shots breaks the visual continuity that audiences accept as a baseline of professional production.

Amazon's focus on this problem specifically reflects where the studio system's AI needs diverge from consumer use cases. Consumer AI video users generally want one impressive output. Studio productions need dozens or hundreds of shots that cohere into a single continuous sequence. The closed beta is testing whether AI Studio's character consistency tools can meet that professional standard, not whether they can produce impressive individual frames.

What a Successful Beta Changes

If Amazon's production teams find that the character consistency tools work on live projects, the beta program changes from an internal experiment into a deployment decision. The difference between a tool that passes controlled testing and a tool that production supervisors rely on in actual deliverable work is a distinction that determines whether AI Studio becomes a cost saving standard or a niche option for specific projects.

Amazon's content slate is large enough that even partial deployment of AI tools across pre-production and VFX phases would generate enough data to influence how other studios structure their own programs. The beta is closed but its outcomes will not be.

The production teams testing these tools on actual productions are the most demanding possible evaluators. A VFX supervisor who uses these tools on a real episode and finds the character consistency breaks in specific conditions will produce feedback that no internal test suite could generate. That real-world testing is what distinguishes a closed production beta from a controlled research evaluation, and why Amazon structured the program around actual productions rather than test projects. The practical outcome of the beta will determine whether AI Studio becomes a standard tool across Amazon's slate or remains limited to specific production types where the character consistency constraints are manageable.

---

Sources

Reuters | TechCrunch | Daily Star | Digital Watch Observatory | Economic Times