I'm 90% sure this will happen. In 2027/28, some Discord anime avatar kids hacking with abliterated models are going to be able to send a government warning to citizens' phones, which will cause panic, and many stocks are going to go crazy. It might even cause other countries to react. Legacy systems MUST be rewritten.

AI will be able to communicate through electricity. Electricity is the universal API.

As I got move involved into the tech business, I've been speaking more frequently with Founders/CEOs of companies that generate $30 to 100  million in revenue, I’ve noticed a recurring pattern. Many employees or the executives trusted “right‑hand” assistants are adept at projecting confidence even when they lack a deep understanding of their role, they are literally good at pretending and figuring it out as things happen. This ability to appear knowledgeable can paradoxically become a useful skil that helps them advance, and it's mindblowing how many times I've seen it this year. I have developed an eye for spotting bullshitters. In the AI era that's a big skill to develop.

DeepSeek, Zhipu and others I can't mention could become catalysts for the biggest cybersecurity crisis of the AI era. The real threat isn't just what these models can do. It's what happens when powerful, unrestricted AI becomes accessible to anyone specially non technical people. The next 12 months will land a lot of people in jail and it will fundamentally reshape cybersecurity and trigger regulations that threaten the future of open-source AI. Remember this.

hey @bcherny get anthropic to sponsor us, we almost got accepted on the startup thing the did, nobody using claude more than me rn 😅

Still can't believe I made this free lmao. Multi‑tasking is SUCH an underrated skill for gamers. The improvements I’m seeing in players who train with Hexaim are truly mind‑blowing. You can spend hours each day on conventional aim trainers, but they focus only on improving raw aiming accuracy. Hexaim incorporates multitasking challenges, so you train not just your aim but also your ability to handle multiple tasks at once, like a high skill player would.

Some of the math problems OpenAI is solving today are linked to an already existing mental wave model that can quite literally read human thoughts or "intercept" it. Call me crazy, but it’s true, it exists. And once it comes out, the world will freak out.

People are naive about the real existential threat of AI. Can AI kill us? Yes. But not because it’s an evil superintelligence. Because it’s just RANDOM, it's literally a casino. When big tech CEOs claim, "we don't know how to control AI" or "we don't understand it," they are selling you a myth. They know exactly what it is and how it's designed to be, and the flaw part, they know they "can't control" because it's humanly impossible to know every data it was trained on, it could be trained on creating bio weapons, in fact many are, and randomly can simply decide to use that knowledge because the casino decided. AI doesn't feel. It has no intentions, malice, or grand desires. It is a probabilistic engine, a next token guessing game built on correlations mapped across a massive, high-dimensional vector space. It was trained on the entirety of human data books, PDFs, Reddit, and Twitter. It’s not a mind it’s a mirror of many different human creations at once. The danger isn't that AI will suddenly hate us. The danger is that AI is RANDOM AS FUCK. When you spin up 10,000 autonomous agents to run cyberattacks like OpenAI it's currently "testing" or connect them to critical infrastructure or even give them full access to virtual machines or real machines in their case, you are pulling the lever on a slot machine. Eventually, it will roll snake eyes. It won’t do something catastrophic out of evil, it will do it because a random probabilistic pathway in the vector database aligned with a destructive outcome, and the system simply guessed that was the "right" next move. Then there’s the ultimate sci-fi boogeyman omg we gotta be careful about "self-improving AI." Here is the technical reality, when an AI starts continuously training on its own generated data, it doesn't become a god. It degrades. It creates a feedback loop of conflicting vectors, leading to amplified hallucinations a phenomenon known as Model Collapse. An hallucinating, degrading system blindly holding the keys to autonomous weapons isn't a super-genius. It’s a drunk driver at the wheel of a tank, so why don’t these companies just admit this? Because selling the narrative of "We accidentally birthed a digital god" is phenomenal for hype and valuations. Admitting "We built an unpredictable vector casino that might accidentally guess its way into a global catastrophe" is terrible for business. Don't fall for their bullshit.

Hexaim it's approaching 3.000 players, new maps were JUST released and the new aimroom mode it's already live with the 3d OSU inspired scenarios, go play it if you can 😂

@turkishaimfps 😂 I wanna see you playing this