Don’t Panic, a London-based creative agency, started with a familiar digital habit: people approve terms and conditions without reading them. Its campaign for the Machine Intelligence Research Institute (MIRI), “Do Not Accept,” turns that habit into an unsettling question about who gets a say in the development of advanced AI.
George McCallum, the agency’s acting executive creative director, and Georgia Stephenson, its associate creative director, describe in written answers to ContentGrip how the idea emerged, why the team rejected science-fiction imagery and how it translated a 90-second film into 15-second subway advertising. Their account also shows the tension between communicating a contested risk and using emotionally direct campaign language.
Table of contents
- How the terms-and-conditions idea emerged
- Why 1,300 public submissions became the emotional center
- What changed in the subway
- How the team handled an alarming claim
- What the campaign asks people to do
How the terms-and-conditions idea emerged
The “Do Not Accept” campaign uses the visual language of digital consent screens to argue that decisions about artificial superintelligence have consequences the public has not been asked to consider. It is an advocacy campaign, reflecting MIRI’s position on advanced-AI risks, rather than a neutral forecast of how AI will develop.
The creative team had explored multiple approaches. Its client was concerned that science-fiction-inspired imagery could make the message sound conspiratorial or theatrical. The team needed a way to describe a possible consequence in language that ordinary people already understand.
The breakthrough came during a discussion about dramatizing a moment of pause. A colleague, Katie, asked whether the campaign could frame superintelligence through terms and conditions, since most people click “accept” without reading the fine print. According to the team, the client immediately embraced the idea.
The metaphor worked for two reasons. Terms and conditions naturally involve what happens if something is agreed to, helping the campaign discuss a future risk rather than depict a certain outcome. And the act of accepting without reading is a behavior most viewers can recognize in themselves.
Why 1,300 public submissions became the emotional center
The film does not rely only on a warning screen. Production partner ProdCo invited people to submit footage of ordinary human moments and received more than 1,300 submissions, according to the creative team. The campaign ultimately used 71 moments from 19 countries.
The filmmakers built short emotional sequences around creation, movement, connection and ageing. Moments included dandelions, spinning, couples kissing and people blowing out birthday candles. These are small personal scenes, not staged visions of an AI future.
The dandelion footage, the team says, was particularly useful. It links the natural world to the human-centered scenes that follow, giving the film a more spontaneous tone. Near the end, birthday candles are extinguished as a warning about human extinction appears on screen. The juxtaposition aims to make the stakes feel personal.
The team’s answer credits director Alex Takács and editor Andrew Litten for shaping the material. It did not describe a participant-permission audit in detail, so the submission total should not be interpreted as proof of a particular licensing or consent process.
What changed in the subway
An audience can give a 90-second hero film enough attention to follow an emotional progression. A commuter passing a digital screen cannot. For New York subway placements, the team reduced the narrative to a few images and a direct question.
The out-of-home treatment used a visual “X” to remove people from the places they occupied, alongside the claim that artificial superintelligence was being built without public consent. It closed by asking people whether they knew the risks. The creative team says transit restrictions on position-based advertisements also pushed the work toward a public-service-announcement style.
That meant changing more than the running time. The film asks viewers to care about the people shown. The subway unit needs to make its point quickly enough for someone to understand while moving through a station. The disappearance effect acts as a bridge between those different formats.
For communications teams, this is an example of designing for the attention an environment permits rather than simply cutting a long film into a shorter version.
How the team handled an alarming claim
The biggest editorial tension lies in the campaign’s wording. Early versions of its most forceful term included “risk of human extinction,” the creative team says. The final wording removed that qualification.
The team says it made the change following developments and commentary that heightened concern about advanced-AI risk. But the possibility and probability of human extinction from advanced AI are disputed subjects, and the materials provided do not establish the prediction as a fact. The wording is best understood as the advocacy campaign’s deliberate choice, not an independently verified outcome.
This distinction matters for marketers tackling difficult public-interest topics. A dramatic line can increase urgency, but it also transfers more responsibility to the evidence and to the context surrounding the claim. The team’s account is useful precisely because it reveals where the trade-off between conditional language and emotional force was made.
What the campaign asks people to do
The campaign directs viewers to the Do Not Accept microsite where MIRI presents its argument and invites visitors to sign their opposition to the development of artificial superintelligence. The creative team says thousands of people had signed at the time it answered ContentGrip’s questions. That figure is self-reported and is not a verified measure of public opinion or campaign persuasion.
The practical action is therefore different from watching the film or sharing a post. The site tries to move an audience from an emotional encounter to reading the organization’s claims and registering a stance. Whether that changes policy or broader attitudes requires separate evidence.
The creative work also drew on the personal interests of its two leads. Georgia curates exhibitions and works with emerging artists, an experience she connects to the public footage call. George creates unusual food photography and recipes, and describes a fascination with making things by hand. Both see collaboration and craft as part of what shaped the finished film.
