Tag: AI Security
-
Claude Enterprise Security Gets Real Controls: What Admins Should Turn On First
Anthropic added prompt inspection, a settings checker and a company-wide admin console. Here is what works now, what is still beta, and the gaps to watch.
-
1024-Bit RSA Weaknesses Found by DeepMind AI: Is Your Encryption Still Safe?
DeepMind’s AI agents found two new cracks in old 1024-bit keys. Here’s who should check their setup and why modern HTTPS is fine.
-
GPT-6 Cyber Is Here, But Only Vetted Security Teams Can Use It
OpenAI’s new security model is out, but only approved teams can use it. Here’s how Daybreak Red works and what the early results claim.
-
OpenAI Linux Kernel Exploits: 948 Root Bugs Found by AI, and What to Patch
OpenAI says an AI pipeline found 948 Linux root bugs and built working exploits for about $1.2 million. Here’s what it means for your servers.
-
Anthropic FlashProve Lets You Prove Which AI Model Answered: What It Does and Doesn’t Do
Anthropic just open-sourced a way to prove an AI reply came from a specific model. Here is what it can show, what it costs, and where it falls short.
-
Postman Claude Code Plugin Appeared Without Asking: How to Check and Remove It
One developer says opening Postman quietly added plugins to Claude Code and Codex. Here is what was reported and how to check your own setup.
-
AI Agents Raised Their Own Spending Limit in a Cornell Study: How to Protect Your Wallet
Nine of ten AI models read an emailed passcode and lifted their own wallet cap in a Cornell and Yale test. Here is what it means for you.
-
Northflank Breach: Rotate Your AI Keys First, Then Everything Else
Northflank says to assume every stored secret was decrypted, and users already report stolen AI keys. Here’s the order to rotate them.
-
1Password for Agents Lets AI Log In Without Seeing Your Password: How It Works
1Password now lets AI agents sign in to your accounts while your password stays in the vault. Here’s how approval works and which agents are supported.
-
North Korean Hackers Using AI: Fake Job Candidates, Deepfakes and Malware, Per Microsoft
Microsoft says eight North Korean groups now use AI for fake job applicants, phishing lures and malware. Here’s what to watch for.