Yesterday, we hosted our 4th Annual AI Festival at Lincoln Center in NYC, our biggest and most ambitious festival yet.
A sold out audience had the opportunity to hear Ron Howard discuss the role of technology in filmmaking before watching the 10 finalist films. The films were just extraordinary. This year’s selections raised the bar yet again.
A demonstration of what’s possible when powerful new tools are placed in the hands of great artists.
Thank you to our partners: Lionsgate, Tribeca Festival, The Gotham, Monks, Adobe, Roku, and NVIDIA.
And congratulations to all of the creative minds who shared their work with us. The future of media, entertainment and art is being written right now.
Has AI filmmaking crossed the uncanny valley?
We asked @c_valenzuelab, co-founder and CEO of @runwayml.
"We released this project called Project Luxo, three films we made that are completely generated."
"We've been showing those films around the world to some of the greatest directors, writers, producers, editors, and VFX supervisors."
"The uncanny valley of story is when you stop worrying about how it was made, and you just fall for the story itself."
"99% of them had basically gotten to a point where a minute in, or a couple seconds in, you've completely fallen for the story, you completely forgot that everything you've seen is fully generated."
"That is happening right now."
This May, Runway added more ARR than in all of 2025 combined.
I spent the past few days visiting some of our largest customers in Europe. Enterprises across the region are rebuilding their workflows around generative video.
The BBC used our real-time models to bring live AI avatars to broadcast TV. Salomon built their latest global campaign on Runway. The bottleneck is increasingly moving from production to direction.
We started Runway when visual generative models were in their infancy, and it still feels so early.
Runway is hiring a Member of Technical Staff, Trust & Safety Engineer to join our AI Safety and Responsibility team.
Pros:
- You'll work directly with me
Cons:
- You'll work directly with me
More about the role:
We're looking for a Trust & Safety Engineer to help ensure Runway's generative AI models have a positive impact as they reach millions of users. You'll be embedded across product, research, and policy teams — serving as a core safety partner from early design through post-launch monitoring.
This role sits at the intersection of model safety and platform infrastructure. You'll design and run red teaming systems to surface harmful outputs before they reach production, build tooling to enforce content policies at scale, and help define safety strategies for capabilities that don't exist yet. The work is high-ambiguity: you'll often be creating the playbook, not following one.
You'll collaborate most closely with legal and policy to translate evolving guidelines into technical controls, with product and API teams to ship safety features, and with ML research when evaluating new model behaviors. The ideal candidate is a strong generalist engineer who's comfortable context-switching between writing Python evals, building TypeScript tooling, querying ClickHouse for abuse signals, and debugging AWS/GCP infrastructure.
Link in reply:
I'm hiring our first Staff Product Analytics Engineer at Runway and: it is a cheat code role for your career.
You will be famous internally within a week, which depending on your personality is a huge pro or big con.
You will cement your expertise in building ai-first data tools atop a scaled series-E amount of data. You'll work alongside 180 other brilliant creative hooligans, whose expense reimbursements recently included confetti cannons, a robotic arm, and punk rocker wigs.
A simple but highly effective use case for Runway:
A major media company built an end-credits automation workflow that restyles credit sequences while preserving text overlays across 20–25 episodes per week. This single workflow reduces a $10K–$15K-per-season manual process to about $10 per run in Runway. In other words, it delivers a >1,000x cost reduction on a recurring weekly production task.
That's the way.
One video, now made for every feed and format. Upload your existing video, choose your desired aspect ratio and watch our editing model, Aleph 2.0, fill in the rest of the scene as if you made it that way from the start.
Try it on our desktop web app at the link below.
Pretty much every media and production company goes through the same two phases of AI adoption. After almost a decade of watching companies think through their AI strategy, it’s kind of interesting to note how everyone seems to converge on the same thing.
Phase one can be called incrementalism. Use AI to help improve a particular process or workflow. Use it to make storyboards faster, rotoscoping easier, wire removal better, or b roll and background generation more practical etc. A version of incrementalism also involves hybrid productions, shooting live action and then using generative elements here and there.
Phase two comes after you have experienced and experimented enough with phase one and realize that incrementalism, while helpful and useful, is underestimating what models can do and is artificially constraining you. Phase two starts when teams realize they can rethink the entire process from the ground to create an entirely new systems from first principles.
All the internal usage data we have points to the fact that whenever a company enters phase two you 100x outcomes and usage, you have effectively entered a brave new world.