AI Manifesto
For four days I wandered the 2026 AI for Good Summit, learning about everything from autonomous trains to agentic education and genetic engineering. Amid the manic pitches for repackaged vaporware, the dry presentations on truly mind-blowing scientific breakthroughs, and the very credible warnings of imminent catastrophe, one thing was in short supply: coherent proposals for using this technology on a practical, day-to-day basis while minimizing harm to society, the environment, and creative communities. So I decided to put together a manifesto of my own, a short collection of rules and principles for AI designed to guide activity at Right Here Right Now, my small videography business in Geneva, Switzerland.
Foreword
In my 10 years as a videographer/director/editor/producer, I have surfed a wave of emerging tech. Drones, compact LED lights as bright as small suns, smart rotoscoping… each advance brought with it a competitive edge in the form of smaller production footprints, easier post-production, or more vivid storytelling. I'd even argue I was among the very early adopters of AI in its various forms, tinkering with prototype diffusion models and enthusiastically embracing the first AI transcription edit workflows.
This time is different. I have never seen any technology evolve at even a fraction of the speed of AI, especially its generative forms. When I hear the endless yammering about how one must adapt or die, I can only push back because I'm not convinced it's humanly possible to stay on top of this progress in any meaningful way. Even as someone used to perpetual tech-scouting and retraining. Even as holder of a Master's degree in Innovation Management. Thus, my heart is slowly preparing for a world where I, and many of the skilled colleagues I love and respect, will have no direct employment in this field.
That said, it's not over until it's over. I still have to earn a living in an industry where AI is an inalienable part of activity. What follows is a first attempt — vague and incomplete as it may be — to navigate this exponential change ethically, and reduce as many negative consequences as possible.
Stay Fit — Avoid Casual AI Use
It's hard — oh so hard — to escape AI entirely these days; its various builds have infiltrated most digital services and even many physical products we interact with. That said, it is possible to limit one's intentional, macro-level use. Skip the LLM query when a simple search will do. Write basic emails by hand. Abstain from summoning silly videos and memes.
Physical exercise keeps our bodies fit. Similarly, critical thinking and creativity require practice to stay sharp. And a population with working capacity in either of these abilities is an essential counterbalance to AI's creeping influence on the world.
Don't Steal Styles
We will never prompt generative AI to replicate the work or style of living artists, whether in writing, visual arts, music, or any other form of intellectual property. While even the act of simply using generative tools is an ethical grey zone because of the use of artistic property as training material and its erosion of demand for human-made work, we believe the most acute form of piracy comes from direct vampirization of proprietary styles. You will never catch us with a Studio Ghibli profile picture.
Don't Appropriate
We will also never prompt AI to depict identifiable communities — ethnic, religious, sexual, or otherwise — so as to avoid perpetuating stereotypes, appropriating cultural capital, or otherwise speaking on their behalf. And, of course, we will never prompt AI to replicate the likeness or voice of any living person without their explicit consent.
Boycott the Baddies
Technology is inherently tangled — economically, legally, infrastructurally — and, as mentioned above, it's debatable whether any AI use can be truly ethical, from either a social or an environmental perspective. AWS provides material support to the Israeli Defence Force, so is using a service based on one of their servers immoral too? That said, some platforms and their ownership stand out when it comes to bad behavior — I think it's only fair to label Elon Musk as persona non grata. His actions through DOGE have directly caused the deaths of hundreds of thousands of people (The Lancet — study on USAID defunding and mortality), and his platform xAI has continued to generate explicit images of minors despite the matter being brought to his attention (CBS News — Grok safeguard lapses involving minors). Boycotting his platforms, and others like them, is the least we can do.
Protect Employment
AI proponents will cite — ad nauseam — how painters were replaced by photographers and horses by automobiles, but they rarely mention the mass suffering that accompanied those transitions. A shift to a “better” techno-economic system, even one with room for humans in it, will bring employment destruction on a truly systemic scale if it follows any historic precedent.
In the absence of any actionable plans for supporting those displaced by AI, we treat any move that decreases demand for human labor as creating a negative externality — much like greenhouse gas emissions. When a task can be done well by a person, we will do our best to hire the person or do it ourselves.
Transparency
If AI is used at any point in our creative workflow, we tell our clients and viewers. We visually label any frames containing more than 30% generated imagery (this threshold is even lower if it involves certain key elements). Clients receive reports about any workflow steps involving AI — a practice we are still formalizing.
This manifesto, for example: the arguments and most of the language are mine (Yes, I actually do use em dashes!). Claude corrected grammar, and proposed completions where I left gaps.
Stay in the Loop
Slop is everywhere, from ridiculously long emails about nothing in particular to weirdly thorough LinkedIn posts with bland literary flourishes, and AI is a big culprit. We don't believe in wasting your time, or in burying true insight and work under mountains of statistical gibberish. Everything we send or publish has been conceived of, mostly created, vetted, read — and understood — by us.
The same principle governs our deliverables: AI may assist — transcribing, rough-cutting, denoising — but it never authors. Every creative decision that reaches a client — framing, pacing, color, story — is made by a human who can explain the reasoning behind it. If we can't defend a choice, it isn't ours, and it doesn't ship.
No Free Renders
We don't run a hundred generations to cherry-pick one, and we factor AI's environmental weight into deciding whether to use it at all.
Trust but Verify
By now, everyone and their brother knows that AI hallucinates. And yet, people continue to share unvetted information from LLMs like kids copying Wikipaedia articles for their homework back in the day. We check every single piece of information before relaying it. There is enough misinformation and disinformation out there without us making it worse.
Exemptions
Coding. If there's a more elegant use of AI than coding and development, I don't know of it. Machines writing for machines. And, given the continued shift of many service providers we rely on towards exploitative business models, using AI to write scripts and internal automation is one way a small enterprise like mine can control costs and remain competitive against much bigger entities. For example, our invoicing software was replaced with a vibe-coded version after they tripled fees for no particular reason.
Legal. Legal language has long been a barrier that only money could cross. Using AI to understand contracts, decode legalese, and draft routine paperwork widens access to rights we already have — provided a professional reviews anything with real stakes.
This is a living document. The technology will change faster than these principles can anticipate, so we revisit and revise them — publicly — as it does. V1.0, July 2026.
Get in touch
Questions about how we work? Tell us about your project.
Contact Us