Jun 30, 2026

Anthropic built a model good enough at finding software flaws that the NSA's own red team couldn't keep it out — then pulled it from the market rather than police who could use it. This week OpenAI announced an equivalent. So did China.

Craig walks Jim through what that actually means for the person listening: if...

Jun 30, 2026

Anthropic shut down its most capable model rather than try to enforce who could use it. OpenAI now has one it says is just as good, likely headed for the same restriction. And China has one too.

Craig lays out the problem that creates for anyone whose job is defense. You cannot protect a network with tools a...

Jun 24, 2026

Anthropic handed its newest model to some of the country's top software companies and asked them to run it against their own code. It found serious security problems in every major piece it examined — not the click-this-link kind, but remote access: connecting to a machine that would normally reject you and taking...

Jun 23, 2026

Craig arrives with bad news and doesn't bury it. Anthropic released a model built to find security problems in software, and gave major U.S. vendors funded access to run it against their own code. It found the serious kind — the flaws where an attacker never needs to log in and takes the machine anyway. Including...

Jun 23, 2026

Anthropic licensed its newest model to software companies so they could review their own code. It found security problems in every single one — including OpenBSD, an operating system audited by human eyes hundreds of times and widely considered the most carefully built in the world. Flaws serious enough that an...