What? Although Mythos was "too powerful for public use" (Anthropic), several Discord users had access to the model from day one! A small group of "unauthorized discord-users" reportedly accessed Anthropic’s powerful Mythos AI model, exploiting a mix of insider access and online
SECURITY
-
Chinese Embassy Censorship and Free Speech in Tech
By
–
I guess this tweet escaped my little Twitter bubble and someone snitch tagged the Chinese embassy like a bunch of CCP Karens calling the fun police. pic.twitter.com/8RufgYUnq2
— 🖤 Christine (@christinelu) 22 avril 2026I guess this tweet escaped my little Twitter bubble and someone snitch tagged the Chinese embassy like a bunch of CCP Karens calling the fun police.
-
LLMs for Permission Control and Secret Blocking
By
–
We been testing llms to judge permissions, and proxies to block secrets. Interesting approach!
-
APIs Limited Releases Business Model Not Safety Policy
By
–
APIs and limited releases for AI models are not a safety policy, they’re a business model (which is totally ok as long as you’re transparent about it). Especially on cyber-security, they give a false impression of control and safety whereas in reality they massively increase the
-
Never Run Applications as Root: Security Best Practice
By
–
There should not ever be a reason to run this as root.
-
CTO Notes on Securing Every Layer of the Vibe Coding Stack
By
–
How we secure every layer of the vibe coding stack- a note from our CTO
-
Anthropic Mythos AI Model Breached by Unauthorized Users
By
–
Big scoop from me: Anthropic's Mythos AI model — the cybersecurity model it says is so powerful it can enable dangerous cyberattacks — is being accessed by unauthorized users. Gift link!
-
Meta CTO No Opt-Out Policy Raises Privacy Ethics Concerns
By
–
“There is no option to opt out” – Meta’s CTO “There was no option to opt out” – epigraph for life as we knew it?
-
Government data compromise: security implications and risks
By
–
At this point, you can safely assume that any data the government has about you is compromised.
-
Three Security Layers for Agents: Isolation, Limitation, and Validation
By
–
Images 2.0 is wildly good at text rendering across even the smallest details The 'thinking' shines with elements the model includes on its own even from basic prompts. "A news broadcast showing Sam Altman meeting with aliens over space data center concerns"