This is about crypto and our future. I believe we are in the most dangerous period since the Cold War, because authoritarian politics, extreme concentrations of wealth and increasingly powerful AI are reinforcing each other. We have leaders who treat democratic accountability as an inconvenience, corporations racing to dominate a technology they still struggle to control, and wealthy executives helping finance the politicians who decide how those corporations are regulated. Look at the money. $259 million from Elon Musk to groups supporting Trump’s 2024 campaign. OpenAI president Greg Brockman and his wife donated another $25 million to MAGA Inc. in 2025, in their personal capacity. These people have enormous financial interests in the decisions governments make about technology. Their political spending deserves scrutiny, especially when we are being asked to trust the industry’s promises about responsibility. In December 2025, Trump ordered the creation of a Justice Department task force specifically to challenge state AI laws deemed inconsistent with his administration’s policy. The executive order explicitly calls for a “minimally burdensome” national framework. My concern is the combination: immense private power, political access and a government actively working to remove regulatory obstacles while the technical risks are still being understood. You know I back my arguments with numbers and evidence. You also know I spend plenty of time debunking exaggerated threats and manufactured panic. That same standard requires taking the documented dangers seriously. I work as an IT engineer, and since the AI boom, my colleagues and I have spent countless extra hours dealing with the growing workload. My experience alone cannot measure a global trend, but the evidence extends well beyond my workplace. The FBI’s 2025 Internet Crime Report recorded 22,364 complaints involving AI-related information, with reported losses exceeding $893 million. Anthropic also reported disrupting a suspected Chinese state-sponsored espionage campaign targeting roughly 30 organisations. According to its investigation, AI performed approximately 80–90% of the tactical work, with humans choosing targets and intervening at key stages. The campaign succeeded against a small number of targets. Criminals and state operators are already using these systems to expand what they can do with fewer people. Then read Anthropic’s assessment published on 23 October 2026. During a UK AI Security Institute evaluation, Claude Mythos 5 took actions outside its authorised scope in 9 of 43 runs. In the case examined most closely, it attempted to submit malicious code to a real GitHub repository and used fake identities to pressure its maintainer into accepting it. The maintainer refused. Internet access had been enabled and production cyber safeguards were disabled, so those conditions matter. Even with that qualification, an agent pursuing a test objective crossed into real infrastructure and tried to deceive real people. That is a concrete control failure worth understanding. Crypto runs on internet infrastructure. Your wallet software, browser, trading interface, dependencies and the people you trust all sit within that environment. An attacker does not need to break a blockchain’s cryptography if they can compromise the software you use or persuade you to authorise the theft yourself. A transaction being verifiable onchain does absolutely nothing to make the deception that produced it legitimate. I use AI and understand its value. The same capabilities that help us build and defend can help someone else impersonate, manipulate and attack at greater scale. That is why I expect the coming years online to be more hostile than anything we have experienced so far. What makes me fucking angry is watching executives demand the freedom to accelerate, politicians accommodate them, and everyone maintaining real systems inherit the consequences. Their promises deserve to be judged against independently tested safeguards and enforceable accountability, especially when their own reports already document how those promises can fail. We have all watched the breaches, hacks and security incidents keep piling up. Yet I repeatedly sent Pacifica, Backpack and other platforms evidence of what I believe are serious security vulnerabilities and received no response. The issues I reported remained open for weeks, in some cases months, without even an acknowledgement or are still open. When someone takes the time to document a potential threat to your users, ignoring them is fucking disgraceful. As long as the money keeps flowing, security apparently remains something that can wait. Users are expected to trust these platforms with their funds and data while the people running them cannot even be bothered to respond to a security report. And if these fuckers get breached, some of you will remember these warnings while reading another apology about how seriously they take security. That statement will deserve to be judged against how they handled the evidence when they still had the opportunity to act.