WAF Weekly: Cisco Live, Palo Alto Under Fire, AI Reshapes Threats

This week, Cisco Live set the tone with a full security overhaul built around agentic AI. Meanwhile, a Palo Alto Networks VPN flaw went from "medium severity" to actively exploited in under a week. AI-generated DDoS testing arrived, and a sophisticated supply chain attack targeted developers through fake tool installers. Here is what WAF operators need to know.

Cisco Live 2026: AgenticOps Platform and Live Protect

Cisco used Cisco Live to unveil Cloud Control, a unified platform that lets human operators and AI agents manage networking, security, compute, and observability from a single login. The AgenticOps vision puts AI agents alongside humans in operational workflows, with humans retaining control over critical decisions.

The security announcements matter most for WAF users. Live Protect now shields more Cisco products from newly discovered vulnerabilities at runtime, with no reboots, no upgrades, and no downtime required. Cisco IQ adds Resilient Infrastructure Services to help customers stay ahead of frontier model risk, plus Quantum Ready Assessments that identify assets most exposed to "harvest now, decrypt later" attacks.

Tom Gillis, SVP of infrastructure and security, noted that AI coding tools have transformed Cisco development internally, but the flip side is that frontier models like Claude Mythos can now find vulnerabilities across entire codebases that humans missed for decades. The patch-to-exploit window is shrinking from weeks to minutes.

WAFplanet angle: Runtime vulnerability shielding is exactly the kind of approach WAF operators need as AI accelerates exploit development. If your firewall vendor cannot respond faster than the attacker, you need additional layers. Solutions like Cloudflare, Imperva, and ModSecurity CRS provide application-layer protection that complements network-level defenses.

Palo Alto GlobalProtect Flaw Exploited Days After Disclosure

CVE-2026-0257, an authentication bypass in Palo Alto Networks GlobalProtect VPN, went from medium-severity advisory to active exploitation in four days. Rapid7 confirmed successful exploitation across multiple customers starting May 17, just days after Palo Alto published fixes on May 13.

Palo Alto initially rated the flaw as medium severity and said it was unaware of any attacks. By May 29, the company had bumped the CVSS score to 7.8 and marked exploit maturity as "attacked." The vulnerability enables credential-less authentication bypass to VPN access, though Rapid7 did not observe lateral movement in the cases it investigated.

WAFplanet angle: VPN gateways and firewalls are supposed to be your perimeter defense. When they become the attack surface, application-layer protection through a WAF is your safety net. Four days from patch to exploitation is the new normal. If you run GlobalProtect, patch immediately. If you cannot patch fast enough, a WAF in front of your applications catches what the compromised perimeter misses.

Palo Alto Earnings Surge as AI Security Demand Accelerates

While one Palo Alto product was getting exploited, the company reported Q3 revenue of $3 billion, up 31% year-over-year. CEO Nikesh Arora declared the "SaaSpocalypse dead for cybersecurity," pushing back on fears that AI would disrupt security vendors.

The numbers behind the demand are striking. Palo Alto fielded 1,200 customer meeting requests in recent weeks from organizations seeking AI security guidance. They completed 800 meetings in 12 weeks, matching their entire previous year total. Arora said customers are no longer asking how to solve today's problems but how to prepare for the next generation of AI-powered threats.

WAFplanet angle: The market is voting with its wallet. AI is making attacks more sophisticated, and organizations are scrambling to upgrade their defenses. This spending surge will ripple across the WAF market as enterprises realize application-layer security is part of the AI defense stack, not separate from it.

MazeBolt Launches RADAR VectorAI for AI-Generated DDoS Testing

Israeli DDoS resilience company MazeBolt launched RADAR VectorAI, a module that uses AI to generate previously unseen DDoS attack vectors and runs them against production environments to expose defense gaps. The company positions it as the "Mythos of DDoS," arguing that defenders need AI-driven validation for network-layer protections.

VectorAI distinguishes between two attack categories: AI-orchestrated attacks (where a model selects and sequences known vectors) and AI-generated attacks (where the model produces novel traffic patterns at query speed). It runs on top of MazeBolt's existing RADAR platform, which simulates DDoS traffic without taking services offline.

WAFplanet angle: DDoS configuration drift is a real problem that static testing misses. If your WAF or DDoS protection provider offers only signature-based detection, AI-generated attack patterns will find the gaps. WAF providers with adaptive rate limiting and behavioral analysis, like Cloudflare, Akamai, and Radware, are better positioned against novel attack patterns.

Fake Developer Tool Installers Hit 88+ Domains via SEO Poisoning

A sophisticated supply chain attack has been running since at least March 2026, using 88+ fake installation pages for popular AI development tools including Claude Code, Cline, and JetBrains. Attackers purchased Google Ads to push fake pages above legitimate documentation in search results, delivering credential-stealing malware that never writes a file to disk.

As of late May, 32 domains were still live. The fake sites were hosted across ten platforms including Squarespace, GitHub Pages, Cloudflare Pages and Workers, Tencent EdgeOne, and Netlify. Some domains redirected visitors to Google searches to artificially boost rankings of other fake pages in the network.

WAFplanet angle: This campaign highlights a growing blind spot. Attackers are abusing legitimate hosting platforms, including ones that offer WAF protection, to distribute malware. The WAF protects the site, not necessarily the visitor. For WAF vendors, this raises uncomfortable questions about how their platforms are used. For developers, verify every download URL against official documentation before installing anything.

WAFplanet Take

The thread connecting every story this week is speed. Cisco is racing to shield products at runtime because patch cycles cannot keep up. Palo Alto's GlobalProtect flaw went from disclosure to exploitation in four days. MazeBolt exists because DDoS configurations drift faster than teams can test them. And attackers spun up 88 fake domains faster than platforms could take them down.

AI is compressing the timeline on both sides. Defenders who rely on manual processes, monthly patch cycles, or "we will get to it" security postures are already behind. A properly configured WAF is not the whole answer, but it is the fastest layer you can deploy between your application and an attacker moving at machine speed.