The grifts are multiplying — and we're still here

Milo Yiannopoulos, the British-born far-right commentator, is sitting in ICE custody in Louisiana after a judge ordered his removal from the United States. He's been held since September 2019, when he was arrested for failing to register with immigration authorities. The original charges centered on a 2017 incident at a UC Berkeley event where he was accused of making threats — though prosecutors later dropped the state case. His detention has been a running story ever since.

He's a UK citizen, which means once the removal order is finalized, he's looking at deportation back to Britain. That's the standard path for noncitizens with final orders of removal. The case has dragged on for years through appeals and legal challenges, which is typical for immigration detentions of this kind.

The broader picture is worth noting: this is one of the longest-running individual immigration detentions in recent memory, and it shows how the system can keep people in limbo while appeals wind through the courts. For the folks who follow this beat, it's a reminder that the machinery doesn't always move fast — and that people can be stuck in it for a very long time.

Why this matters for us: the same system that holds one person for years is the same system that pulls working families apart, and the delay between arrest and removal is where most of the damage gets done.

The business model everyone expected to be free is suddenly the most valuable thing in the room. Open-weight AI companies — the ones that publish models for anyone to use — are drawing serious acquisition bids from the big players. The money is flowing to the people giving models away.

Why? Because the big labs want the underlying tech without the liability. Open-weight lets anyone run the model on their own hardware. It's the kind of thing a restaurant can't use a trademark on, and a big AI firm can't easily shut down. Buying the company is how they get the weights, the training data, and the engineering team — without the baggage of a closed API they'd have to maintain.

This isn't a one-off. The pattern is setting up a new market layer: companies that build open models, get acquired, and then quietly disappear into a bigger stack. The models keep getting better, the prices keep dropping, and the capital is flowing to the open side of the equation.

Why this matters for us: the tools that power the apps we use every day are going to be owned by fewer and fewer companies — so we need to pay attention to who controls the open ones, because that's where the actual tech lives.

Neocloud Lambda just closed a $1 billion private-debt round to stock up on Nvidia AI chips. The plan is straightforward: buy the hardware, lease it to Microsoft, and collect the spread. It's one of many of these loans in a row — the kind of deal that signals how much capital is chasing the same hardware right now.

The AI buildout has turned data centers into a financial market. You can't just buy chips anymore; you have to borrow against them, and the lenders are taking the risk. Each new loan is a bet that the demand for compute will keep rising fast enough to cover the interest. The real question is whether the spread between what Neocloud pays for the chips and what Microsoft pays to use them stays wide enough to service the debt.

Why this matters for us: when the money for chips flows through private loans instead of public markets, the risk hides until it doesn't — and the people who get hit are the workers and neighborhoods that bet everything on these booms.

A new neuroimaging study flips the usual story about psychedelics: psilocybin doesn't disconnect people from the world — it does the opposite. Brain activity becomes more tightly coupled to what's actually happening around them.

For decades, the cultural script has been that these substances send you off into your own head. The data is catching up to something else entirely. The brain's networks, rather than shutting down or going off the rails, lock on to external stimuli with more precision than normal.

Why this matters for us: if the old fear-mongering about psychedelics is wrong, then the real question is whether the church, the bodega owner, and the abuela who swears by san pedro are onto something we just stopped listening to.

A new survey shows more people oppose police use of license plate readers than support them — a reversal from years when the technology was largely unchallenged. The backlash has been building as reports pile up about police using the cameras for purposes that go well beyond the original justification of tracking stolen vehicles.

What's happening is this: the cameras are everywhere — at police stations, on patrol cars, at checkpoints — and they log plates continuously, creating searchable databases of where everyone has been. Once the data is in, it can be used for everything from tracking protests to profiling communities. The optics are clear: if you're a Brown family driving past a checkpoint in a neighborhood that's already over-policed, this isn't about your stolen car. It's about being watched.

Why this matters for us: these cameras are one of the most invisible forms of surveillance in our streets, and the fact that a majority now sees through the cover story means the conversation is shifting in our favor — we should push hard to ban them before the next wave of police tech rolls in.