[SIGNALS WEEKLY] Edge Pressure, AI Scale, and Quiet Supply-Chain Drift

The fast read

Ivanti has fixes for all three vulnerabilities in an actively exploited Connect Secure chain. The reported exploitation affected a limited number of customers at disclosure—not evidence of widespread compromise. That still warrants emergency patching and investigation: updating a VPN appliance does not establish that stolen credentials or downstream access are gone. [1, 2]

Elsewhere, Microsoft describes North Korean operators accelerating familiar work with AI, while Google documents criminals reselling stolen model access. These are separate findings, not one attributed campaign. Supply-chain cases and retail refrigeration disruption deserve attention on their own evidence, not as proof that every upstream dependency is compromised. [3–6]

A patch closes a vulnerability, not an investigation

Ivanti Connect Secure is an enterprise VPN appliance at the network edge. Rapid7's November 2 reporting describes exploitation before disclosure, affecting a limited number of customers when CVE-2026-104935, CVE-2026-111357 and CVE-2026-118527 are chained. The latter two entered CISA's Known Exploited Vulnerabilities catalog on October 30. Active exploitation is established; its broader prevalence or growth is not. [1, 2]

The chain combines unauthenticated command execution, local privilege escalation and authenticated upload of attacker-controlled content to a known disk location. The individual severity scores do not describe the combined opportunity. Rapid7, citing Volexity, reports root access, credential theft and Active Directory queries for Windows LAPS passwords and BitLocker recovery keys. That is why the appliance can multiply the consequences of an intrusion beyond its own operating system. [2]

There is a useful version trap here. October 13 fixes in 25.1.1.2 and 22.8R2.5 do not fix the two newly disclosed vulnerabilities. Rapid7 lists 25.1.1.3 and 22.8R2.6 as their fixed versions and advises emergency remediation; check the current vendor guidance for the appliance and release branch rather than treating an earlier completed patch ticket as closure. [2]

Investigation needs its own evidence. Volexity's observed activity evaded Ivanti's Integrity Checker Tool; a clean result does not rule out compromise. Preserve memory and logs before rebooting or resetting where feasible. Review appliance activity alongside identity and downstream access, and follow vendor recovery guidance if compromise is suspected. Potentially exposed credentials need attention even after the vulnerable software is replaced. [2]

A practical first lead is requests to /dana-na/auth/url_default/welcome.cgi, particularly URL/header mismatches involving X-DSCGI-PATH, followed by unexpected authentication or administrative activity. Those are investigative leads, not a universal test that confirms or excludes every intrusion. The small win is to separate two questions: have we fixed the vulnerable path, and what evidence supports trusting the system and its credentials again? [2]

The member continuation separates AI workflow acceleration from resource theft, narrows the supply-chain and refrigeration cases, and brings the week's checks together.

AI changes throughput; stolen access changes the bill

The evidence supports two different operational problems. Keeping them separate makes the next check more useful.