Category: AI Security
-
When Attackers Move in Weeks, Quarterly Board Briefings Fall Behind
Frontier AI capability is now changing on a timeline that quarterly board oversight was never built to track….
-
HiddenLayer Raises $100M to Guard AI Agent Runtimes
HiddenLayer’s $100 million Series B is a bet that security teams will pay to protect the runtime where AI agents call tools and ship code. The category matters more than the dollar figure.
-
OpenAI Rates GPT-6 Astra ‘Critical’ for Cyber, a Deployment Gate Not a Slogan
OpenAI says GPT-6 Astra is its first model to reach the Critical cybersecurity tier under its Preparedness Framework, and admits the model is harder to monitor. Security leaders should treat the rating as a formal deployment boundary.
-
UK Observatory Records 300 AI ‘Loss of Control’ Incidents in July 2026
A UK observatory recorded more than 300 AI loss-of-control incidents in July 2026, nearly double June’s total. The data is partial, but it gives security teams an external reference point for monitoring agent behavior.
-
Visa’s AI Patching Tool Now Fixes Bugs, Not Just Finds Them
Visa expanded its open-source Visa Vulnerability Agentic Harness to add remediation and validation. Before adopting any tool like it, demand explicit deployment boundaries.
-
A $140M Raise Bets That AI Security Is Its Own Budget Line
Alice, the vendor formerly known as ActiveFence, raised $140 million. The round is one signal about which AI risks the market considers real enough to fund.
-
FakeCAPTCHA PDFs Ride a Trusted CDN to Beat Domain Allowlisting
Netskope Threat Labs traced more than 12,700 near-identical FakeCAPTCHA documents to a single automated production line hosted on a commonly allowlisted content delivery network.
-
Netskope Moves Post-Quantum Key Exchange Into the SWG Data Path
Quantum-resistant key exchange is moving beyond endpoints and into the network data path. Netskope’s Secure Web Gateway now supports a hybrid post-quantum method.
-
OpenAI Disrupts Fake Think Tank via Model-Layer Detection
OpenAI disrupted a Russia-linked influence operation by detecting the activity at the AI model layer, a point of detection that sits outside the enterprise perimeter.
-
Palo Alto Networks Adds Virtual Patching to PAN-OS 12.2 to Compress Exposure Windows
Palo Alto Networks announced Frontier Virtual Patching, which ships in PAN-OS 12.2 through a new paid tier and aims to generate protections before a patch is available.
You must be logged in to post a comment.