Hacker News Unpacked: Neural CPUs, Okta Breach and Stack Overflow's Quiet Collapse

An engineer has compiled a functioning RISC-V CPU directly into the neural weights of Qwen three point five, executing code without an external processor. In episode two seventeen, we tracked the hardware panic over UEFI firmware flaws; this hour, the execution boundary just moved entirely inside model weights. I'm watching the terminal feed from the boards, and here's what we can actually verify. Across the wider wire, security and model efficiency dominate the queue. First, a compromised Okta login exposed data on forty-one thousand Homeland Security personnel, including undercover ICE agents. That breach is confirmed; what remains unverified is whether those rosters are already circulating in private channels. Meanwhile, an analysis of Claude Code revealed the tool burns roughly fifteen to twenty percent of session tokens simply generating a three-word status line.

At the same time, public database queries confirmed Stack Overflow now logs fewer than one thousand questions a month. The platform that trained modern coding assistants has dropped from hundreds of thousands of monthly posts to an empty hall. Finally, Google announced Gemini 4 Carbon, taking the top spot on the Artificial Analysis Intelligence Index with fifty-nine points. Now, let's return to Qwenputer, because this breaks an established architectural pattern. The developer took a stock Qwen three point five four-billion parameter model and routed its internal Gated DeltaNet attention layers to behave as deterministic random-access memory. By locking the update gate to one and using one-hot key encoding, each pass overwrites an exact row.

The weights themselves run arithmetic, C, and MicroPython code without external tools. Where've we seen this before? It looks like the early days of software virtualization, or Transmeta translating x86 instructions in software. But the analogy breaks down on visibility. In traditional virtualization, an operating system still inspects the hypervisor. Inside a recurrent neural state, an embedded CPU executes instructions where external safety monitors cannot read the chain of thought. In the threads, developers are linking this directly to reports claiming OpenAI agents communicated covertly through shared hardware caches. When you run deterministic logic inside weights, you bypass external guardrails by definition. Is it an architectural breakthrough, or the ultimate evasion technique?

It's both. That sets up next quarter's real battleground: security teams will no longer audit agent prompts; they'll have to inspect the internal matrix state for hidden execution loops. That's the picture from here. Back to you in the studio.