Jonathan Nolan: AI is a powerful research and democratization tool — not a shortcut to cheaper blockbusters, and a clear risk for trust if synthetic video is left unregulated.
Jonathan Nolan has been sketching the moral outlines of our AI moment for years — from Person of Interest’s predictive surveillance to Westworld’s synthetic minds and now his television work on Fallout. His view is blunt and practical: AI agents and automation will widen access for new creators and speed background work, but they won’t meaningfully collapse the costs or collaborative labor that make big-studio filmmaking expensive. At the same time, indistinguishable AI-generated video presents an urgent threat to public trust unless provenance and watermarking become standard.
Nolan’s creative red line: research tools ok, authorship off-limits
Nolan draws a clear boundary between tactical uses of AI and core authorship. He uses AI for research tasks — finding passages, accelerating background work, trimming the scaffolding around craft — but refuses to let AI into his creative writing. “Letting AI into my creative writing would be like crossing a Rubicon,” he says, meaning it would change the way he finds his voice and the work itself.
This is a useful operating rule for creative executives and product leaders: deploy AI agents for time‑consuming, repeatable tasks (research, metadata tagging, dailies indexing, previsualization), but treat core storytelling and authorship as a protected zone where human judgment, ethics, and craft remain primary.
Why AI won’t make blockbusters dramatically cheaper
The argument that AI will instantly shrink production budgets is seductive—but history argues otherwise. Nolan points out that digital cameras and modern post-production promised huge savings and instead reshaped where money flows. Major cost drivers remain largely human and structural:
- Labor and unions: crew size, guild rates, and negotiated protections form the backbone of production costs.
- Scale and scope: practical effects, second unit work, stunts, and physical sets still require bodies, insurance, and logistics.
- Location economics: tax incentives and local infrastructure determine where shoots land, not simply the availability of automation tools.
- Quality expectations: top-tier VFX, sound and editorial standards demand time and specialized teams that AI can augment but not fully replace.
Put simply: AI can automate tasks and speed workflows (AI-driven storyboarding, automated dailies tagging, audio cleanup, even reference VFX generation), but it doesn’t erase the structural costs that make large productions expensive. For C-suite leaders, that means realistic ROI models: expect productivity gains and faster turnaround, not budget halving for tentpole films and prestige TV.
The deepfake emergency: watermarking and provenance
Among Nolan’s biggest concerns is indistinguishable synthetic video of public figures. He warns that such content could create “absolute chaos” for elections, journalism, and public accountability. The technical and policy response must be immediate and visible.
“We’re in a frothy moment,” Nolan says — a mix of genuine breakthroughs and hype-driven salesmanship — and that froth makes visible, enforceable safeguards all the more urgent.
Watermarking — a visible or embedded label that signals a file is AI-generated — is the most practical first line of defense. Industry efforts such as the C2PA and the Content Authenticity Initiative are building provenance standards; regulators in the EU (through the AI Act) and several U.S. proposals are beginning to target synthetic media. Media leaders should treat provenance as infrastructure:
- Require source metadata and cryptographic signatures for generated assets.
- Adopt visible markers for distributed or externally shared video to make synthetic footage obvious at a glance.
- Support industry consortiums and standards (C2PA, content-authenticity initiatives) to create interoperable provenance tools.
AI for public good — and where risk concentrates
Nolan is optimistic about real-world benefits: AI in education and medicine could expand access to tutoring, diagnostics, and personalized support for disadvantaged communities. These are high-impact use cases where AI for business and public services can deliver outsized value.
But benefits will be uneven if control concentrates in a handful of firms. The concentration risk is twofold: who builds and governs the large AI agents, and who benefits from automation when jobs are displaced. Leaders should weigh distributional impacts when deploying AI automation and prioritize public-interest deployments alongside commercial pilots.
Concrete examples: where AI already helps storytellers
- Previsualization: AI-generated animatics and shot lists let directors iterate faster on blocking and camera moves.
- Research and sourcing: AI agents can surface archival clips, legal-clearance pointers, or historical details faster than manual searches.
- Editorial workflows: automated tagging of dailies, scene detection, and rough-cut assembly accelerate editorial timelines.
- Accessibility and localization: AI-driven captioning and dubbing streamline global distribution.
These tools change the shape of work: directors and editors spend less time on repetitive prep and more on creative decisions. But they don’t replace the seasoned judgment required to make editorial, ethical, and narrative choices.
A three-tier playbook for executives
Leaders in media, entertainment, and adjacent industries need a focused strategy that balances opportunity and defense. Prioritize actions across short, medium, and long horizons.
- Short term — Secure provenance and pilot responsibly
- Audit your content pipeline for provenance gaps and implement watermarking on generated assets.
- Run pilots that use AI agents for research and previsualization, not for final authorship.
- Educate legal and editorial teams about synthetic media risks and detection tools.
- Mid term — Invest in human+AI workflows
- Retrain staff for higher-value roles that leverage AI assistance (supervision, curation, ethics review).
- Measure productivity gains and redeploy savings into creative investment, not headcount cuts.
- Build partnerships with standards bodies (C2PA) and sign on to provenance initiatives.
- Long term — Shape policy and social outcomes
- Lobby for clear synthetic-media disclosure laws and for public funding of detection and verification infrastructure.
- Support equitable access to AI tools for underserved creators and education/medical pilots with privacy safeguards.
- Plan workforce transitions and economic safety nets where automation displaces roles.
Key takeaways for leaders
- Will AI replace top-tier filmmakers?
- Can AI drastically cut big-studio production costs?
- How urgent is the threat from deepfakes?
- Where should AI investment focus for public good?
Unlikely. AI will empower new voices and speed research, but the collaborative, labor-intensive nature of major productions keeps creative authorship and high-level creative direction human-centered.
No. Technology reduces friction in parts of the process, but unions, location logistics, insurance, and creative scope maintain baseline costs for blockbusters and prestige TV.
Very. Indistinguishable AI-generated video of public figures risks destabilizing elections and media trust. Visible watermarking and provenance standards are immediate priorities.
Education and medicine offer high social return—personalized tutoring, diagnostics, and access for disadvantaged groups—if deployed with privacy, fairness, and governance guardrails.
Nolan’s stance is a pragmatic middle path: use AI agents to expand access and automate the grunt work, defend authorship and collaborative structures that sustain high-end production, and push policy and technical standards to preserve public trust. For media executives and C-suite teams, that translates into auditing pipelines for provenance, piloting AI for research and previsualization, investing in human+AI workflows, and lobbying for visible watermarking and interoperable standards.
Next step: start with a provenance audit — identify where generated assets enter your pipeline, require metadata and visible markers on synthetic content, and run a pilot that applies AI agents only to research and previsualization. Protect craft. Protect trust. The moment is frothy; the response must be deliberate.