In partnership with

Anthropic's top skeptic spent two weeks trying to prove its most powerful model was overhyped. It wasn't — and what he found could protect your business.

Nicholas Carlini's whole job is to doubt AI. So when he pointed Anthropic's most powerful model at the Linux kernel — the code running most of the internet — he expected noise. Instead, one AI surfaced 479 real security bugs in software he'd never cracked himself, plus a critical flaw in Ghost that attackers were exploiting within weeks. That same push is now hardening the code behind Apple, Microsoft, Google, JPMorgan and roughly 200 other organizations.

The full story — how the "Carlini Loop" works, and the one simple habit that keeps your tools safe no matter what — is on the site.

Quick Bites

Anthropic's two most powerful models went dark — A federal export-control order pulled Fable 5 and Mythos 5. Whatever happens next, the lesson is permanent: never single-source the AI your work runs on — keep a fallback like Opus 4.8 one setting away.

The Carlini Loop has become a cult document — His talk demonstrating the technique has racked up hundreds of thousands of views in security circles.

Claude Security brought it to the rest of us — The public-beta scanner reads your codebase and suggests patches for review, putting the same bug-hunting in every team's hands.

The full breakdown — how the loop works, the Ghost-bug timeline, and what it means for your stack — is on the site.

Read the full story →

Free to read · no signup

Help make better ads

Did you recently see an ad for Roku Ads Manager in a newsletter? We’re running a short brand lift survey to understand what’s actually breaking through (and what’s not).

It takes about 20 seconds, the questions are super easy, and your feedback directly helps us improve how we show up in the newsletters you read and love.

If you’ve got a few moments, we’d really appreciate your insight.

JOIN 46,000+ PROS

Get the AI edge, every morning.

Was this forwarded to you? Get one short, useful AI briefing — free, 5–6 mornings a week.

Subscribe free →

Free forever · unsubscribe anytime

## What kind of content do you want to see more of

Help me shape the future of AI Super Simplified! Vote below to let me know what you want to see more of.💬 Don't see your top choice? Just hit "Reply" to this email or drop a line straight to [email protected]. We read every single message and l

Login or Subscribe to participate

If you enjoyed this issue and want more like it, subscribe to the newsletter.

Brought to you by Stoneyard.com  •  Subscribe  •  Forward  •  Archive

Keep Reading