Fortinet patches nearly 100 vulnerabilities, including about 17 critical, after adding frontier AI ...

17 of the nearly 100 newly disclosed flaws are critical and nearly 80 are high-severity; Fortinet did not say any are exploited in the wild [1] .

The release includes fixes from Fortinet's frontier AI-assisted testing program, and the company thanked Anthropic, OpenAI, Google and SpaceX for collaborating [2] .

FortiMail took the largest of critical fixes, including a path traversal and an HA command-injection bug, both rated CVSS 9.8 [1] [3] [4] .

Fortinet disclosed nearly 100 vulnerabilities across its product line on Nov. 10, a batch it says includes flaws surfaced by frontier AI models it has folded into its product security testing [1] [2] . 17 are critical, several allowing arbitrary code execution, and nearly 80 are high-severity, SecurityWeek reported [1] . Fortinet made no mention of any being exploited in the wild [1] .

The advisories follow a coordinated "security hardening release" Fortinet shipped in October, when it pushed fixed builds of FortiMail and FortiAuthenticator and then, on Oct. 22, FortiOS, FortiManager and FortiAnalyzer while holding back the details until this month's Patch Tuesday [2] . "Fortinet would like to thank Anthropic, OpenAI, Google, and SpaceX for their collaboration throughout this process," the company wrote, saying the models "analyze code paths, protocol handling, and configuration logic at a depth and speed that traditional fuzzing and manual review cannot match" [2] .

Where the critical bugs are

FortiMail got the largest batch of critical fixes, followed by FortiAuthenticator, FortiOS, FortiManager and FortiAnalyzer, with FortiSASE and FortiClient updates also resolving critical issues [1] . In FortiMail, two flaws rated CVSS 9.8, a path traversal and a command injection via crafted HA join-request packets, let an unauthenticated attacker execute code or commands [3] [4] ; others allow takeover of IBE recipient accounts and encrypted webmail sessions [1] . FortiManager has an improper-authorization flaw in its FGFM protocol that lets an unauthenticated attacker who can reach the FGFM port run code [5] . Baldur Security's Kevin Joensen said a FortiClient bug he calls Fortinet2Shell, exploitable when FortiPAM is enabled, stems from a CORS misconfiguration that lets a malicious web page reach a local service and "reliably execute arbitrary code" [1] .

Fortinet told customers to treat the fixes as "a priority change rather than a routine maintenance cycle," warning that AI helps attackers diff patches into working exploits "in hours or days, rather than weeks" and chain lower-severity bugs [2] . The batch is separate from CVE-2026-104286, a FortiMail flaw that CISA and Fortinet flagged as exploited in the wild in early October [6] .

The volume fits a wider pattern this month: Microsoft's November Patch Tuesday fixed 1,093 flaws, including an exploited Outlook bug credited to a researcher using Anthropic's Claude ( MLQ.ai coverage ), and SAP shipped four maximum-severity fixes ( MLQ.ai coverage ).