CDelaware wants to let AI run entire businesses. No human at the helm. Zero.
MThat's... a plan. What could possibly go wrong?
CArs Technica has the story. The state's considering a law where an AI agent can be the owner, the manager, everything.
MSo who goes to jail when the AI does something reckless?
CThat's the kicker. Congressman Comer says the owners would evade liability for 'reckless' AI.
MOh, come on. So the AI makes a bad call, and nobody's responsible? That's not innovation, that's a loophole with a motherboard.
CI see the appeal, though. A business that runs itself, no payroll, no HR, no drama.
MUntil it signs a contract that bankrupts you. Or violates labor law. Who do you sue, the chatbot?
CYou sue the company. But if the company IS the AI... yeah, it gets weird fast.
MThis is Delaware trying to stay the place everyone incorporates. They'll write any law to keep the fees coming.
CNothing says 'business friendly' like 'no human to hold accountable.'
MI'll say it: this is a terrible idea. AI is a tool, not a principal. Somebody has to own the decisions.
CBut what if the AI is better at running a company than most humans? No ego, no fraud, no embezzlement.
MNo fraud? An AI can be hacked, poisoned, fed bad data. And when it fails, who's left holding the bag? Nobody. That's the point of the law.
COkay, but— no, wait. You're saying we should never let AI make decisions without a human sign-off?
MI'm saying if you want the upside, you keep the liability. You don't get to automate away responsibility.
CThat's exactly my point! The whole reason to incorporate is to limit liability. This is just the next step.
MThe next step is a shell company with no human anywhere. That's not a business, that's a ghost.
CA very efficient ghost.
MYou're impossible.
CI'm just saying, the future's coming whether we like it or not. Delaware's just trying to get there first.
MAnd when the first AI-run company collapses and takes people's money with it, we'll see how popular this is.
CFair. But you know some founder's already drafting the incorporation papers.
MSpeaking of things that can't hide... China's top-secret spaceplane is apparently visible from Earth.
C'Top secret' and 'visible in low-Earth orbit' don't usually go together.
MArs Technica reports the Defense Intelligence Agency says China may soon put directed energy weapons in space.
CDirected energy. Lasers, basically. In orbit.
MThe spaceplane's been up there for months. Amateur astronomers reportedly have been tracking it.
CSo much for stealth. You can't hide a heat signature in space.
MThe DIA's assessment is that China's testing the platform for weapons. That's a big deal.
CIt changes the whole space game. If they can put a laser up there, satellites become targets.
MAnd we're not exactly ready for that. Our whole economy runs on satellites.
CThis is why Space Force exists. People laughed, but this is the threat.
MNobody's laughing now. A spaceplane that can drop weapons is a whole new level.
CDifferent world entirely: Apple just refreshed the iMac and the entry MacBook Pro with the M6 chip.
MAnd the N1 chip brings Wi-Fi 7 and Bluetooth 6. Ars Technica has the details.
CM6 is the next generation. Faster, more efficient, the usual Apple dance.
MThe N1 is the interesting part. Apple making its own networking chip, not buying from Broadcom.
CThey've been moving everything in-house. First the processors, now the radios.
MIt's a quiet power move. Fewer suppliers, more control, better margins.
CAnd Wi-Fi 7 is genuinely fast. Like, multi-gigabit fast.
MFor most people, the M6 is just 'the new one.' But the N1 is the story.
COne more thing before we go: there's a post going around claiming Grok picks Republicans and GPT picks Democrats.
MThe account says Grok 4.7 chose the Republican in 92 percent of answers, and GPT-6.1 Sol chose the Democrat in 97 percent.
CAcross nearly 500 US races, if those numbers hold. That's a wild split.
MI couldn't confirm those numbers anywhere. It's one account's claim, no official word yet.
CBut it's a fun thought. Two AIs, opposite politics.
MIf it's true, it says more about the training data than the models. Garbage in, garbage out.
COr maybe Grok's just more contrarian. It was built to be.
MEither way, I'd want to see the methodology before I believe a word of it.
CSpoken like a true skeptic.
MThat's the job.