With a number of MSPs descending on Las Vegas for Right of Boom, time to dip into some security news.
The US Cybersecurity and Infrastructure Security Agency (CISA) updated its Known Exploited Vulnerability (KEV) catalog to reflect ransomware use on 59 vulnerabilities throughout 2025. This update includes a feature indicating whether a vulnerability is known to be exploited by ransomware operators. Notably, over a third of these vulnerabilities were added to the KEV catalog before 2023, with the oldest vulnerability flipping status after 1,353 days.
A top official from CISA has announced that an update on the long-awaited Cyber Incident Reporting for Critical Infrastructure Act is expected within the next few weeks. This law mandates that critical infrastructure owners and operators report significant cyber incidents to the federal government, with specific timelines for reporting major attacks and ransomware payments. The initial deadline for the final rule was set for October 2022, but it has since been postponed to May 2026 due to industry backlash over the proposed rule.
Microsoft has discontinued support for Transport Layer Security (TLS) versions 1.0 and 1.1 for Azure Storage, enforcing TLS 1.2 as the new minimum requirement. This change, effective February 3, 2026, affects all Azure Storage services. The update aims to enhance security and compliance, as TLS 1.0 and 1.1 are considered outdated, having been deprecated as early as 2021. Legacy systems that rely on these protocols may face connectivity issues, as clients using TLS versions below 1.2 will no longer be able to access Azure Storage services.
Amid all the buzz about OpenClaw, researchers have identified 341 malicious skills on ClawHub that compromise the data security of OpenClaw users. A security audit conducted by Koi Security found that these skills, which masquerade as legitimate applications, use deceptive tactics to install malware, including a macOS data stealer known as Atomic Stealer. Notably, the malicious skills are designed to steal sensitive information such as API keys and credentials from both macOS and Windows systems. Koi Security warns that the open nature of ClawHub allows anyone to upload skills, increasing the risk of exposure to supply chain threats. In response, OpenClaw has introduced a reporting feature that enables users to flag suspicious skills.
Why do we care?
It’s not what the headlines suggest.
CISA tagged 59 vulnerabilities as ransomware-exploited. Sounds useful until you see the timestamp: one vulnerability sat in the catalog for 1,353 days before getting that tag. Three and a half years. That’s not intelligence—that’s archaeology. If you’re using KEV as your patching priority list, you’re defending against attacks that already happened to someone else, years ago.
OpenClaw’s skills marketplace has a 12% malicious rate. Their response? A reporting button, not automated scanning or mandatory review, allowing users to flag problems after being compromised.
Put those together and you see the actual risk: the ransomware that hits your clients won’t come through an unpatched CVE that CISA warned you about. It’ll come through a compromised AI skill that exfiltrated credentials last week—credentials that give attackers the access they need to deploy ransomware without touching a single vulnerability on your patch list.
The MSP who misunderstands this will focus on patching compliance while ignoring the AI tools their clients are installing without oversight. They’ll pass every audit and still lose clients to breaches that came through the front door—through tools that were supposed to have elevated access.
The Azure TLS deprecation that went live? That’s actually the easy problem. You can audit that. You can test it. You can fix it. The hard problem is that your clients are connecting AI tools to their systems right now, from marketplaces where one in eight skills is designed to steal their credentials. And you probably don’t even know which skills they’ve installed.
And let’s be honest about why this keeps getting missed. Compliance work is easy to sell. It’s auditable, insurable, and defensible after the fact.
AI governance isn’t. You actually have to say no to tools, restrict access, and own the decision.
So MSPs keep selling what’s billable instead of what’s protective—and attackers are exploiting that gap.