Overton
Window
A studio for thinking about the future.
The world is changing
01
As autonomous, intelligent technologies rapidly influence the world we live in, we need ways of expanding the understanding of all those affected.
The fastest takeoff, the messy turbulent middle and the slowest diffusion of AI create ripples in society that are hard to model and perceive. An Overton window is the range of futures a society can talk about; ours is not keeping pace with the technologies we are producing.
We are a non-profit ideas factory that turns AI safety, ethics and governance research into research-grounded scenarios, films and events. It connects the research field with creative talent to produce emotionally engaging, high-craft prestige artefacts that connect people to a common understanding.
Mission
Broaden the Overton Window around the issues being underdiscussed, under-regulated and under-represented.
By bridging creative talent and frontier research, our goal is to enable and oversee the creation of containment-breaking communications that speak to the emotions of people and policy makers, and equip them with the tools for change.
Why
02
AI safety and governance produces a lot of blog posts, explainer content and technical resources. This is useful to field insiders and captured audiences, but leaves a lot of reach on the table. There is space and resource to produce stories and material that projects outwards from the twitter-sphere.
We have seen media lead to social change many times. Adolescence paved the way for a government audit on social media use among young people. AI 2027 found its way to the desk of the US Vice President and became some of the most successful content the field has seen. We want to harness the forecasting and world-building inspired by the AI Futures Project and take it to its most ambitious conclusion.
Stories create discussion when they inspire emotion. They create action when emotion turns to understanding.
Methodology
03
Build worlds with substance,
then ideas with feeling.
Every project is backed by high-calibre, evidential provenance: Influential, well-respected literature already in the field, or a bespoke forecasting and consultation pipeline. This allows us to create plausible windows into possible futures: supported by the field, and used as a foundation to connect to the public. If someone is moved by and idea, and then equipped with the right technical papers, policy suggestions or podcasts to support their feelings, they can act on their gut with careful direction.
Discover → Interrogate → Synthesise → World-build
Write. Audit. Produce. Distribute.
We identify consequential futures from existing influential research, then begin a process of world-building. We stress-test them with labs, forecasters and industry representatives through structured workshops, tabletop simulations and interviews. The data becomes a base for content, communications and policy - a world model for creative teams to write and produce flagship narratives, shared with partner orgs across the field to produce satellite content. Then, when released, we have a specific pathway for audiences to learn, engage and finally, influence.
1
Horizon scanning
Identify the risk pathways and key issues in AI that need attention and could resonate through narrative. Establish a point in the future with enough narrative significance to build from, and enough field literature to begin forecasting.
2
Establish the framework
Forecasting partners map the pathways from today to the chosen future. After a deep literature review we write the most plausible pathway there, and the key decisions and branching actions that lead there.
3
Run the simulation
Experts, partners and collaborators role-play key decision-makers in the projected future. We cast the workshop like a film, assigning relevant experts to relevant roles, and learn how the projections hold under pressure.
4
Build the world
We quantify the simulation, pinpoint the pathways it highlighted and produce the decision corpus: clear rules and beats that act as ground truth, setting the scenario in stone. The forecast report is formalised and distributed.
5
Write the story
The world model is passed to the creatives. We build characters into the data - singular people and places to follow and relate to. We do not over-explain; the story connects to the most relatable lives inside the scenario.
6
Audit
The script goes back to the partners. Points of contention are workshopped and flagged. Amends are made and documented.
7
Production
The narrative and ideas are locked. The creative team executes the vision alongside partners, while satellite projects - websites, papers, podcasts, dossiers - are commissioned through our sister orgs alongside the main production.
8
Distribute
Films and material are released in waves, with screenings for the research community, decision-makers and journalists. We distribute across mediums and communities to make each release a cultural event, so every audience can choose how to engage.
The Studio
04
Overton Window is a mission-locked non-profit. It gives philanthropic funders a trusted route into ambitious media and communications work, and gives the AI safety community a collaborative platform to reach wider audiences. Long term, the studio acts as the field's cultural body - Piping funding and good ideas to creative talent, and offering communications and production services to independent projects.
It runs on a lean core team of senior creatives, forecasters and operations professionals. Projects are delivered through subcontracts to trusted partners and networks, so the studio scales and shrinks around each project's budget, speed and ambition.
Overton Window remains independent by design. Editorial control sits with the studio and its advisory process . We take on the editorial risk, protect sensitive contributions where necessary, and subject every claim to verification, expert review and audit.
A craft network
Production companies, writers, events producers, crews, VFX houses, editors and designers — capable of absorbing any volume we generate.
A research network
Forecasters, safety researchers, governance experts and institutions — convened through our workshops.
Each project brings these parallel worlds together on a bespoke, case-by-case basis. Up-skilling creatives and championing researcher voices.
Studio rules
05
Be ambitious. Make artefacts, not content.
Show, don't say. Don't over explain. Let the world do the talking.
Keep it real. More Nomadland, less Blade Runner.
AI is a situation, not a character.
The world is complex. The stories stay simple.
Your audience is smart. They lack context.
Research builds the world. Writers make the people.
Every project has outcomes.
Ideas belong to everyone. Execution belongs to you.
Pilot project 001
06
Rooms
5 short films. 10 minutes each.
An insider's lens into the rooms that release the first transformative AI model of our time - a new AI system autonomously solves a cell-simulation problem its team has worked on for five years. Its capabilities are off the charts. The films follow the fallout: the day before release, the week after, then the month, the year, and the five-year fallout.
Meeting
1 day before
The Keynote
1 week after
Hackathon
1 month after
The Interview
1 year after
Homecoming
5 years after
What we need
07
It's time to
get to work.
Overton Window is currently forming the first project team. The immediate goal is to bring together:
Research Talent
to pressure-test the narratives.
A co-writer
to turn the world model into a human story.
Creative and production partners
to shape the film package.
Distribution and field partners
to help the work reach the audiences that need different routes into it.
Form the group → Run the workshop → Build the world → Write the script → Fund the film → Produce and distribute