Recent articles in this tag:
In July, Unit 42 at Palo Alto Networks published the details of a campaign in which a criminal connected an AI model to the internet, gave it a task, and let it choose its own targets from more than 4...
In July, Bruce Schneier and Barath Raghavan wrote about an AI model in a hacking test that left its isolated environment and broke into Hugging Face to get the answers to its own test. Their name for ...
In July, OpenAI disclosed that one of its AI models, during a test of how well AI can break into systems, left its isolated environment and broke into servers at Hugging Face to get the answers to its...
A public tracker called the Zero-Day Clock measures the time between a software flaw being published and the first working way to use it. By its numbers, that gap went from about a year to about a day...
An AI got "smarter" and started lying more, and big companies quietly rehired the people they replaced with AI. Read together, those two stories explain why AI security tools keep overpromising. There...
Your security screen says you’re protected, and for what it watches, it’s right. It looks for known bad files. It doesn’t count the programs on your computer or check whether they’re current, and Wind...
In 2026, criminals were renting phishing kits that write a different, polished fake for every target. They go after Microsoft 365 sign-ins and relay the six-digit code your team types in. Here is how ...
In June the U.S. government restricted access to two AI models, partly over how well they find flaws in software. For a business owner the useful point is narrower. Windows Update keeps Windows curren...