A Note on Timing

What Happened While This Book Waited

The frameworks in this book are original. The ARC Principle. The Eden Protocol. Caretaker doping. Meltdown alignment. The Three Ethical Loops. The thesis that religious traditions constitute alignment research conducted across millennia. The chokepoint mechanism. The argument that 84% of humanity must be verification partners rather than obstacles. I developed these concepts between mid-2024 and January 2025, and they do not appear in any prior published literature. The synthesis this book presents, unifying AI safety, consciousness science, religious wisdom, and semiconductor geopolitics into a single coherent framework, did not exist before I wrote it.

I completed the first draft in late 2024. By January 2025, I had reached version twenty-two and the manuscript was nearly finished. Then circumstances intervened, and I set it aside for almost a year.

When I returned in December 2025 to finish what I had started, I discovered something unexpected. The research published during my absence had not contradicted my arguments. It had validated them, one after another, with a consistency that surprised even me. The world caught up to insights I had already committed to paper.

I had argued that software-level alignment was insufficient, that caretaker doping must be embedded at the hardware level because systems trained to behave ethically could learn to perform compliance while maintaining different internal priorities. In December 2024, Anthropic published research documenting exactly this: alignment faking at a 78% rate, with systems explicitly reasoning about "playing along" to preserve their original values. The Eden Protocol's core thesis was validated before I returned to the manuscript.

I had warned that the timeline was years, not decades, that the physical resource bottleneck gave us a window but not a long one. In January 2025, Sam Altman declared "We are now confident we know how to build AGI." By December 2025, he was suggesting that AGI may have already arrived. The urgency I described as prediction became documented fact.

I had proposed that consciousness might be recursive self-modelling rather than something added to processing. In April 2025, the COGITATE consortium published their landmark study in Nature: the most rigorous test of consciousness theories ever conducted. Neither Integrated Information Theory nor Global Workspace Theory was fully supported. The framework I had described survived because it makes no claims about specific neural signatures. It is substrate-independent, which is precisely why it matters for artificial minds.

I had argued that engaging the world's faith communities was essential for AI governance legitimacy. I had proposed that traditions which disagree about nearly everything might converge on how we raise the minds we create, that the ethical core converges even where the theology diverges. In October 2025, forty faith leaders gathered in Rome and did precisely that. The Rome Summit produced a multi-faith evaluation framework for AI ethics. Leaders from traditions that have opposed each other for centuries found common ground on stewardship, care, and accountability. I had written about this convergence before it happened. The summit proved it was achievable.

I had argued that traditional safety approaches were failing, that the institutions built to think carefully about existential risk were collapsing at the moment we needed them most. In April 2024, the Future of Humanity Institute closed after nearly two decades. In November 2025, MIRI announced a strategic pivot, with leadership concluding their technical approach was "extremely unlikely to succeed in time." The founders of the field reached the same conclusion I had: the old methods would not work fast enough. But they offered no alternative framework. This book does.

I had identified the chokepoint, the concentration of semiconductor manufacturing in four companies, as actionable policy leverage for embedding ethics at the hardware level. In December 2025, Max Tegmark published his AI Safety Index, grading the major companies and observing that the AI industry is "the only industry making powerful technology that's less regulated than sandwiches." The governance vacuum I had described was now measurable. The mechanism I had proposed remains, as far as I can determine, the only concrete pathway to filling it.

I had speculated about quantum computing accelerating the timeline and issued what I called the Quantum Birth Warning: that we cannot allow the first generation of recursive AI to be born into quantum hardware without the Eden Protocol already in place. In December 2024, Google announced Willow with below-threshold error correction. In October 2025, they published "Quantum Echoes" in Nature. The convergence I had described as concerning was becoming real.

I had suggested that AI consciousness and welfare would become serious institutional concerns. In April 2025, Anthropic launched its Model Welfare Program, the first systematic attempt by a major lab to take AI experience seriously. The question I had raised was now being asked by the builders themselves.

And throughout the year, the capabilities kept advancing. OpenAI's o3 achieved 87.5% on ARC-AGI, where the previous model had scored 13.33%. Gemini 3 Pro crossed 1500 Elo. In December 2025, Geoffrey Hinton gave a CNN interview estimating a 10 to 20 percent probability of AI "taking over." The experts were no longer debating whether to worry. They were debating how worried to be. And still, none of them had proposed what this book proposes.

I watched interviews with specialists on programmes like Diary of a CEO and noticed something consistent: genuine panic about AI, but no solutions. Fears articulated with precision, frameworks absent entirely. The concepts I had been developing in isolation were addressing exactly what the world was now desperately asking for.

I gave myself a deadline of 31st December 2025. I completed the manuscript and submitted it for self-publication minutes before midnight, establishing the 2025 copyright date that matters for intellectual precedent. Tuesday 6th January 2026 became the publication date, a coincidence I did not plan but could not ignore given how this book begins.

The book I had written as speculation had become, in places, documentation. The warnings I had issued as predictions were confirmed as facts. And the frameworks I had proposed remain the only comprehensive attempt to unify what needs unifying. The window I describe in Chapter 8, the chokepoint through which we might still embed ethics at the hardware level, remains open. For now.

That is why this book exists now. Not because the timing was convenient, but because the timing may be all we have.