AI's eating the world and the engineers are tired

Anthropic is pulling the plug on two of its models, Fable and Mythos, after a Trump administration directive. The models are going dark, and the move signals how fast federal shifts can reshape the AI landscape.

This isn't just corporate housekeeping. Fable and Mythos have been workhorses for Anthropic's product line, and their shutdown ripples through the companies and teams that built on top of them. Workers, partners, and the developers who ship tools powered by these models are now recalibrating.

Why this matters for us: la migra's reach into tech means even AI companies aren't immune to the new administration's pressure — and the ripple effects hit the side hustles and small teams that depend on stable tooling.

The UK is weighing a wide-ranging ban on social media for children under 16, following Australia's lead. The proposal would block access to a broad swath of platforms for anyone in that age bracket, moving beyond the age-verification rules that have been the norm.

Australia passed its own version of the law last year, and the UK is now looking at something similar — though the details are still being worked out. The government is weighing whether to enforce the ban through platform design (like restricting what under-16s can see or who they can follow) or through stricter age gates at the door.

This is part of a growing push to treat social media less like an open playground and more like a regulated space. Australia's law has been the template; other countries are watching closely to see if it holds up in practice. The question is whether the UK will go further or just copy the playbook.

Why this matters for us: Our kids grow up on these platforms too, and if the UK sets a precedent, it could reshape how social media works for young people — and how parents hold platforms accountable.

Cockroach Labs published a look at how companies are managing costs when running AI agents at scale. The post walks through the infrastructure challenges that show up when you move from a prototype to production — compute, storage, and the kind of operational overhead that sneaks up on teams.

The company's own data is the backbone here. They're looking at what happens when you run AI agents across regions, handle retries, and keep latency down under load. The numbers point to a practical question: how much of your spend is actually moving work forward versus burning through tokens and connections just to keep things alive.

Why this matters for us: as AI agents move from hype to the daily grind, the companies that figure out cost control first will be the ones we actually trust to run our systems reliably.

Scientists at the Innovative Genomics Institute have developed a CRISPR technique that selectively shreds cancer cells while leaving healthy cells intact. The method works by targeting specific markers on tumor cells and triggering a cascade that tears them apart from the inside out.

The breakthrough matters because previous CRISPR cancer approaches often went after healthy cells too — causing collateral damage in the body. This new version is more precise, which means fewer side effects and potentially broader use. Researchers have tested it in animal models so far, with results showing the technique can clear tumors without wrecking surrounding tissue.

Why this matters for us: Cancer doesn't care where you come from, and treatments that work better and hurt less help our families get through it without losing everything to the process.

The Linux Foundation announced OpenSharing, a project to standardize how AI models and datasets move between different platforms and tools. The initiative lays out common formats for sharing model weights, training data, and metadata so organizations aren't stuck playing whack-a-mole with proprietary formats.

The push comes as the AI industry has become a patchwork of incompatible standards. Companies building models on one platform often struggle to move them to another. OpenSharing aims to cut through that friction by establishing shared rules for how assets are packaged, described, and exchanged — less about inventing something new, more about agreeing on what already works.

Why this matters for us: when AI data exchange gets standardized, the cost of building and sharing tools drops, which means smaller teams and independent developers can compete with big tech instead of being locked into their formats.

Researchers have built a new CRISPR technique that selectively shreds cancer cells while sparing healthy tissue. The old approach was blunt — it cut DNA everywhere, which meant collateral damage to the cells we actually need.

The new method uses a molecular tag that only activates in cancer cells, so the gene-editing scissors stay sheathed until they reach their target. Think of it like a key that only fits one lock. The result: fewer side effects, tighter targeting, and a path toward treatments that hit hard without wrecking the house.

Why this matters for us: CRISPR has been promising for decades, but the side effects have been real — and when treatments hurt the people they're meant to heal, la gente stops trusting them. This one gets closer to working like it should.