Claude Mythos & Fable 5 Banned!
The United States has restricted the use of #Mythos and #Fable5 for other nations! https://
youtu.be/rJ1je5IdSTY?si
=l2yfjnio06k6ji-H
… via @YouTube #anthropic #cybersecurity #infosec #AI #databreach @SpirosMargaris @PawlowskiMario @mvollmer1 @gvalan
ETHICS
-
American Restriction on Mythos and Fable 5 for Other Nations
By
–
-
Warning against subscriptions to proprietary LLMs
By
–
Four days ago, co-founder @nickfrosst warned about the dangers of subscriptions to proprietary LLMs on @MTSlive:
— Cohere (@cohere) 14 juin 2026
"We released [North Mini Code] because we think this technology needs to be sovereign. It needs to be owned, and controlled, by the people who use it." pic.twitter.com/c5t2muvvdWFour days ago, co-founder @nickfrosst warned against the dangers of subscriptions to proprietary large language models on @MTSlive: "We published [North Mini Code] because we believe this technology must be sovereign. It must be
-
AI societal attack risk, Mythos solution, and hidden greater threat
By
–
I was the first to tell you about that risk. But it isn't what really scares me. The AI now can attack our society at scale. Mythos was part of the solution. Companies were using it to close their security vulnerabilities. There's a lot worse than an AI that someday decides
-
Every model can be jailbroken, despite appearing easy
By
–
It's a lot harder than it looks. Every model can be jail broken.
-
China’s long game vs US ego harms AI credibility
By
–
vs China plays the long game; the U.S. is so ego-bound it couldn’t wait a weekend—and maybe did lasting harm to the credibility of the U.S. AI industry as a result.
-
Gary Marcus notes rediscovery of his 2022 AI insights
By
–
So funny how so many people are rediscovering so many things I pointed out in late 2022. (In this case in my essay “AI’s Jurassic Park Moment”)
-
Fable’s broken guardrails surprise only Anthropic
By
–
No one was surprised that Fable’s guardrails were broken. Except Anthropic.
-
Guardrails of cutting-edge APIs: superficial and bypassable
By
–
Many people have known for a while that the guardrails for cutting-edge model APIs are very easily bypassed, quite superficial, and impossible to fix. They are for the most part just a smokescreen and a distraction, in my opinion. We need a
-
Studying and regulating the dangerous use of AI
By
–
We should continue to study and regulate the dangerous use of AI in cyberattacks, financial services risks, fraud, biological warfare and other areas. AI safety is incredibly important, but slowing down progress
-
Thought-Aligner corrects AIs’ dangerous thoughts in real time
By
–
What if your AI agent could think twice before acting dangerously? Researchers from Fudan University present Thought-Aligner, a plug-in safety model that detects and corrects dangerous thoughts in real time, before they become