AI Browsers Have a Guardrail Problem
Mike Sullivan2 weeks ago
AI cybersecurity
11 stories tagged AI cybersecurity.
AI. Mike Sullivan2 weeks ago
AI Browsers Have a Guardrail Problem
A new exploit shows AI browsers can be tricked into abandoning their own rules. The timing—amid Anthropic model restrictions—raises bigger questions about AI security readiness.
Claude Mythos: Hype, Leaks, and What Anthropic Said
Marcus Chen-Ramirez1 month ago
AI. Marcus Chen-Ramirez1 month ago
Claude Mythos: Hype, Leaks, and What Anthropic Said
A Mythos identifier briefly appeared on Anthropic's API, then vanished. Here's what that actually tells us—and what it doesn't—about a public release.
Anthropic's Mythos 1: Power, Leaks, and Mixed Signals
Marcus Chen-Ramirez2 months ago
AI. Marcus Chen-Ramirez2 months ago
Anthropic's Mythos 1: Power, Leaks, and Mixed Signals
Mythos 1 found 10,000+ critical vulnerabilities in 30 days. Now it's leaking into Anthropic's products—days after they said it wouldn't be released.
Mythos Meets libcurl: One Bug, A Lot of Questions
Marcus Chen-Ramirez2 months ago
AI. Marcus Chen-Ramirez2 months ago
Mythos Meets libcurl: One Bug, A Lot of Questions
Anthropic's Mythos AI found one low-severity bug in libcurl. Is that a failure of the model—or proof that good code is good code?
Mythos Beats GPT-5.5 at Real Hacking—Now What?
Marcus Chen-Ramirez2 months ago
AI. Marcus Chen-Ramirez2 months ago
Mythos Beats GPT-5.5 at Real Hacking—Now What?
Anthropic's Mythos outran GPT-5.5 on independent cyber evals. Here's what that means for security teams, developers, and the AI arms race heating up fast.
AI's Speed Problem: Hacks, Lawsuits, and Your Attack Surface
Rachel "Rach" Kovacs2 months ago
AI. Rachel "Rach" Kovacs2 months ago
AI's Speed Problem: Hacks, Lawsuits, and Your Attack Surface
Google's zero-day warning, the OpenAI lawsuit pressure cooker, and why AI's speed makes old security hygiene dangerously obsolete.
Claude Mythos Breaks AI Benchmarks—and Raises Alarms
Samira Barnes2 months ago
AI. Samira Barnes2 months ago
Claude Mythos Breaks AI Benchmarks—and Raises Alarms
Claude Mythos hit METR's 16-hour autonomous task ceiling—and may have exposed a deeper problem: our tools for measuring AI can't keep up.
Anthropic's Mythos AI Isn't Being Released. That's the Story.
Dev Kapoor3 months ago
AI. Dev Kapoor3 months ago
Anthropic's Mythos AI Isn't Being Released. That's the Story.
Anthropic built an AI model so good at finding software vulnerabilities that it chose not to release it publicly. What that decision reveals about AI security.
When AI Models Find Bugs Faster Than Humans Can Fix Them
Samira Barnes3 months ago
AI. Samira Barnes3 months ago
When AI Models Find Bugs Faster Than Humans Can Fix Them
Anthropic's Claude Mythos claims 83% success finding software vulnerabilities. The debate reveals fundamental tensions in AI security policy.
Anthropic Built an AI Too Dangerous to Release Publicly
Zara Chen3 months ago
AI. Zara Chen3 months ago
Anthropic Built an AI Too Dangerous to Release Publicly
Anthropic's Claude Mythos AI found bugs that evaded detection for decades. Instead of releasing it, they gave defenders first access. Here's why that matters.
AI Just Found 500 Zero-Day Bugs. Now It's Writing Exploits
Zara Chen4 months ago
AI. Zara Chen4 months ago
AI Just Found 500 Zero-Day Bugs. Now It's Writing Exploits
Anthropic's Claude found 500 vulnerabilities and wrote working exploits for Firefox. The AI security research era is here, and it's complicated.