The Superintelligence Mirage: Why the AI Futures Project Report Ignores the Decentralization Imperative

Over the past seven days, the AI Futures Project’s report on superintelligence has been paraded as a break in the clouds—a call for US-China cooperation that promises to reshape global AI policy by 2040. But any risk consultant worth their salt knows to look past the vision board. The report’s core assumption—that two centralized superpowers can collaboratively build a safe superintelligence without decentralized infrastructure—is a mathematical impossibility. I spent last week reverse-engineering the report’s implicit trust model, and what I found was a governance framework built on sand. The ledger remembers what the marketing forgets: transparency isn’t a diplomatic gesture, it’s a cryptographic requirement. The report, published by the AI Futures Project, outlines a cooperative roadmap for achieving superintelligence by 2040, emphasizing shared safety standards and joint research. It paints a picture where the US and China pool resources to avoid a catastrophic arms race. On the surface, it sounds rational. The industry hype cycle has long treated superintelligence as the ultimate prize, and any call for de-escalation is initially welcomed. But the devil is in the data layer. The report offers no technical verification mechanism—no on-chain proof of compliance, no auditable ledger for research contributions, and no decentralized storage for safety protocols. It assumes trust between sovereign entities, which in the current geopolitical climate is like assuming a 50% APY yield is sustainable. Trace every byte back to the genesis block: this report is a whitepaper without a working testnet. Here is the core of my teardown: the report’s vision relies on centralized governance—a small group of policymakers and lab directors making decisions about alignment, data sharing, and shutdown procedures. Based on my forensic audit of the Imperfect Finance protocol in 2020, I learned that any system with a single point of failure will inevitably be exploited. In that case, a centralized reward distribution algorithm caused 40% dilution within six months because no one could verify the math on-chain. Similarly, the AI Futures Project’s cooperation framework lacks a decentralized verification layer. Without it, either side can cheat—alter training data, hide research, or embed backdoors—without detection. The report treats security as a policy choice, not a technical invariant. Code does not lie, but developers do. The only way to prove that both parties are adhering to safety protocols is to put every commitment on an immutable blockchain, with time-stamped hashes and zero-knowledge proofs. The report doesn’t mention a single smart contract. Let’s go deeper. The report implicitly assumes that superintelligence’s control problem can be solved through mutual agreement. This is naive. During my forensic analysis of the FTX collapse in 2022, I traced 1.2 billion USDC through Alameda wallets and proved that solvency was a mathematical illusion created by a centralized ledger. The AI Futures Project’s cooperation is the same: a centralized ledger of promises with no public validation. The report claims that shared safety standards will prevent a race to the bottom, but without an on-chain oracle that monitors real-time compute usage or training distributions, how can either side verify that the other isn’t secretly scaling an unaligned model? The answer is they can’t. Metadata is not ownership; it is merely a pointer. What the report calls a “cooperation agreement” is actually a set of pointers to goodwill. Pointers can be redirected. Moreover, the report completely ignores the role of decentralized compute marketplaces and on-chain governance in mitigating superintelligence risk. In 2026, I audited a prominent AI trading protocol that promised autonomous profitability. The AI was simply reading centralized news APIs and executing trades—its decisions were opaque, unverifiable, and easily manipulated. That protocol collapsed when I exposed the centralized oracle dependency. The same risk multiplies at the superintelligence scale: if the AI’s core training or inference runs on a centralized cloud, the owner has a kill switch that can be abused. The report’s cooperation model would concentrate that kill switch in a bilateral committee—two centralized points of failure instead of one. Greed optimizes for yield, not for survival. The only way to ensure that superintelligence benefits all of humanity is to decentralize its control: distribute compute across multiple jurisdictions on a blockchain, enforce safety constraints via smart contracts, and make every training iteration verifiable through zk-SNARKs. The report shies away from this because it threatens the very power structures it was designed to preserve. But the contrarian angle: the report does get one thing right. It explicitly names superintelligence as an existential risk and argues that going it alone (either country) is dangerous. That is a step forward from the usual techno-optimist denial. The report’s authors correctly identify that no single corporation or nation can solve alignment alone. Where they err is in assuming that two nations are better than one. The math of game theory suggests any bilateral oligopoly tends toward collusion or defection—neither is stable. The real solution is a multi-stakeholder, permissionless verification layer. The report’s “cooperation” is, in effect, a centralized cartel. A mirror reflects the face, not the value. The report reflects the face of diplomacy, but it doesn’t reflect the value of verifiability. Its blind spot is the assumption that trust can be built through meetings and agreements, rather than through cryptographic proofs that anyone can audit. The takeaway is stark: without a decentralized, on-chain governance framework for superintelligence development, the AI Futures Project report is just another piece of narrative. It will be forgotten the moment a real crisis hits—when someone deploys an unaligned model that no committee can stop. Risk is a number until it becomes a breach. The breach is coming. The only question is whether we will have a transparent, immutable ledger to trace its origins, or whether we will be left guessing as the system melts down. The choice is between trust and truth. Truth lives on-chain.