Anthropic restricts Claude Mythos AI release after model exposes software vulnerabilities
Updated
Updated · The New York Times · Apr 28
Anthropic restricts Claude Mythos AI release after model exposes software vulnerabilities
14 articles · Updated · The New York Times · Apr 28
Anthropic has limited access to its new Claude Mythos AI model primarily to large companies due to its ability to uncover previously unknown software vulnerabilities.
This restriction aims to give major organizations time to secure their systems, but leaves smaller companies, nonprofits, and individuals more exposed to cybersecurity risks.
Experts warn that advanced AI tools like Claude Mythos could empower attackers and increase threats to everyday users, urging immediate improvements in personal cybersecurity practices such as strong passwords and regular software updates.
With AI hackers emerging, are small businesses facing an unwinnable cyber war?
Should powerful 'hacker AI' be regulated like a weapon under international law?
As developers use AI to write code, are they creating a new wave of vulnerabilities?
If AI can find any software flaw in minutes, is our digital world fundamentally broken?
Passkeys were meant to kill phishing. Can criminals already bypass this new defense?
Is Anthropic's 'Project Glasswing' creating a new class of digital haves and have-nots?