CYBR.SEC.Media

Featured

AI-Generated Code Still Fails 38% of Security Tests Despite Gains, Veracode Finds

Veracode's 2026 GenAI Code Security Report finds AI coding security improving for the first time, but persistent injection flaws and weak Java performance demand stronger security testing. (Includes infographic)

The AI Apocalypse Can Wait: What [un]prompted II Revealed About Cybersecurity's Future

Bill Brenner

Leading security researchers confronted AI-driven disruption at [un]prompted II, finding reasons for optimism amid the threats to hacking careers.

AI Won't Kill Your Hacking Career. Being An AI Luddite Will.

Bill Brenner

This week: All the news from [un]prompted II, flaws emerge in 7 AI platforms, and help comes to besieged rural hospitals.

Community Corner

See all

From the CYBR.SEC.Community

AI in Media

On The Use of AI-Generated Images

CYBR.SEC.Media uses AI-generated images instead of generic stock photos. Here's how we try to do it with transparency.

, and Bill Brenner

· October 30, 2026

Above the Packet: Field Notes from the Meaning Layer

Own Your World Model

A kilometer of fiber, two houses, some very heavy machinery, and the world cybersecurity is supposed to protect.

, and Chris Blask

· October 26, 2026

AI and Vulnerability Management

AI Is Finding Old Vulnerabilities Faster Than We Can Fix Them

LLM-assisted bug hunting is uncovering critical vulnerabilities in old software such as FATFS, exposing a growing cybersecurity problem: AI is accelerating vulnerability discovery while patching and remediation struggle to keep pace.

, and Michael Farnum

· October 19, 2026

How to CISO

Crypto Agility: Preparing Encryption for Quantum Threats

As quantum computers become cryptographically relevant, agentic networks will have to keep one step ahead of the decryption storm.

, and Andy Ellis

· October 16, 2026

Adventures in Threat Hunting

Building Determinism in an AI World: Why AI Agents Need Proof of Control

AI agents operate unpredictably, creating security blind spots that traditional controls can't address. The emerging Proof of Control standard aims to make agent actions auditable, traceable and independently verifiable.

, and Michael Smith

· October 13, 2026

Latest Articles

See all

AI innovation and riskRisk Management

AI Innovation Requires Risk. Cybersecurity Can't Be the Department of No.

, and Michael Farnum

As businesses accelerate AI adoption, security leaders risk becoming obstacles to innovation. It's time to rethink how cybersecurity enables risk-taking.

October 29, 2026

[un]prompted IIAI Governance

Shadow AI Exposed: How One Company Fixed Its Broken AI Governance Process

, and Bill Brenner

A pharmaceutical company discovered 35 unauthorized AI tools despite having a formal approval process. At [un]prompted II, two former CISOs revealed how a redesign quadrupled AI requests and slashed approval times.

October 28, 2026

[un]prompted IIAI self-healing software

AI Agents and Self-Healing Software: What's Next

, and Bill Brenner

At [un]prompted II, Gadi Evron explored AI agents that detect vulnerabilities, patch live software and verify code before deployment.

October 28, 2026

[un]prompted IIAI replacing hacker skillsAI hacking

Gadi Evron on AI Replacing Hacker Skills: “Art Didn’t End With the Brush”

, and Bill Brenner

At [un]prompted II, Gadi Evron reflects on AI replacing hacker skills, cybersecurity jobs and why the hacking profession must reinvent itself.

October 28, 2026

AI and Vulnerability ManagementPrompt Injection FlawsEndor Labs

Prompt Injection Flaws Expose 7 AI Platforms to Remote Code Execution

, and Bill Brenner

Endor Labs researcher Peyton Kennedy uncovered 14 vulnerabilities across seven AI platforms, exposing paths from prompt injection to remote code execution and data theft. (Sponsored by Endor Labs)

October 28, 2026

Agentic AIAccess ControlEndor Labs

AI Agents Are Getting More Access Than Security Teams Can Control

, and Bill Brenner

As enterprises rush to deploy AI agents, Endor Labs researcher Peyton Kennedy warns that excessive privileges, weak isolation and misplaced trust are expanding the attack surface. (Sponsored by Endor Labs)

October 28, 2026

Critical PHP VulnerabilitiesPHP2ShellAI and Vulnerability Management

Certificate Authorities, PHP Runtimes, and the Limits of Defense-in-Depth in the “Vulnpocolypse” Era

, and George V. Hulme

The eMudhra vulnerability exposed a years-long window of significant certificate risk. The PHP2Shell vulnerability is ubiquitous. Last week’s five vulnerabilities targeted many of the security controls assumed to be trustworthy.

October 27, 2026