Anyhere Software

Prepare a draft treaty or multi-lateral agreement to pause frontier AI model development until such time that it can be proven to be safe and there is global agreement to proceed. There need to be methods to verify compliance and all major countries must participate.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

Our regulatory bodies are falling dangerously behind the capabilities and threats represented by current AI models, let alone ones likely to come over the horizon in the coming months and years. By the METR organization most cited by AI researchers, the growth curve is accelerating exponentially with a 4-7 month doubling of capabilities. We must stop this race to superintelligence by a handful of corporations. To neglect this situation is to surrender human sovereignty without discussion.

Incorporating AI into our systems of governance and operation creates dependencies that will not serve humanity in the long run, even if they seem like a good idea to many world and business leaders. Once we are dependent on AI, we may not be able to pull the plug, even if we had the ability and reasons to, because it would mean a different sort of catastrophe. We are children playing with fire, and it will burn everything. Experts in the field, even leaders of the major AI corporations, have told us as much.

I have been a programmer for over 40 years, and I have never seen anything like the transformation currently taking place in my sector. Software engineers are helping the AI industry replace first junior then middle then senior programmers in the U.S. and across the globe There is no plan, only wild hopes for profit driving this shift. Greed will seal the fate of one job after another. The AI heads may think they hold the reins, but no one knows what ASI will be like, only that it will have the ultimate say over our fate. Is this what we want? No one voted for this, and if the multinationals have their way, no one will ever have that opportunity.

We need groundswell to counteract the business case for AI and human replacement. This dialog is essential to a considered decision how to proceed with new technology. Past crises have included nuclear proliferation, global famine, and pandemics. The current crisis is rapid AI developments, but future technologies will present similar problems, and we need global dialog and governance to weather our future as a human society.

Pause AI is the group I participate in, and they have global reach in promoting legislation and treaties through non-violent political discourse and action. We are attempting to amplify what many experts are saying of the potential harms of the current unbridled race to superintelligence. Our best hope is for the U.N. and/or major players to come together in agreement for a pause to enable us to weigh the risks posed by current AI and neutralize the threat of ASI to life on our planet.

The first step is to take seriously the AI developers' own warnings about the technology they are producing. Claude Mythos was not released by Anthropic because it presented a clear threat to global informatics security, but it seems it has already been leaked by the one or more of the partners it shared it with in Project Glasswing. OpenAI just released ChatGPT 5.5 with similar capabilities and apparently no safeguards against abuse. These companies must be held liable for the damage likely to come of their carelessness.

So far, there is clear consensus from scientists and technology experts inside and outside the AI community that this is an existential threat to humanity, and possibly to all life on Earth. Unheard and underrepresented are the people likely to lose their jobs to AI in the next few years. If robotics manage to catch up, this will include many skilled tradeworkers across multiple industries. There is no plan for this, other than vague suggestions that a universal basic income will be needed. I shouldn't have to point out that this has never been tried anywhere, ever, and is unlikely to work. Worse, it leaves all the wealth and decision-making to a handful of unelected individuals who have thus far shown nothing better than a rampant disregard for their fellow man.

At this point, a global public service advertisement push is needed to inform people of the coming storm before it hits. We have super-cars racing around on a track without guardrails, and starting to think maybe we need some. We need to stop the race, first! We can't build guardrails during an active race, any more than we can install flood protection during an actual hurricane.

First and foremost, we need a pause on frontier AI model training. We can start by halting the build-out of data centers throughout the world. We have an AI economic bubble driving an ill-considered race to ever more capable and general models. This needs to stop, and a global treaty is the best (and perhaps the only) way to achieve this. The data centers are a cost to the environment we cannot afford, driving a race that may lead to our own extinction. This is really a no-brainer if ever there was one. Verification that a pause is being honored by every country is obviously essential. Once this is done, we need to look at regulations on how AI is used, how jobs are protected, how children and the vulnerable are protected, and what our rights are regarding the fruits of our labor. AI companies have trained their models on stolen data. Reparations are in order at this point, as well as controls to provent such theft moving forward.