Malvertising, Trusted Tools, Real-Time Attacks & Shrinking Windows

This week’s conversation focuses on a shift in how initial access is actually happening: attackers aren’t just breaking in, they’re being gaining access through tools, workflows, and actions that already look legitimate.

Across this week’s headlines, that shows up in a few ways. Fake AI developer tools delivering infostealers through install commands. Malicious NPM packages turning normal dependency use into a supply chain risk. Real-time phishing through fake Zoom calls that lead directly to remote access.

At the same time, the window between exposure and exploitation is collapsing from days to mere hours with the advancements in AI, i.e., a firewall 0-day moves straight into ransomware activity. Industrial systems designed to stay isolated are now reachable and, in some cases, already being disrupted.

Trust and speed remain the throughline, but they’re showing up in more places and moving faster than most teams are prepared for. Attackers are operating inside normal activity, and by the time it looks suspicious, access is already established.

Key Takeaways:

Infostealers are being disguised as Claude Code, OpenClaw and other AI developer tools, TechRadar

https://www.techradar.com/pro/security/infostealers-are-being-disguised-as-claude-code-openclaw-and-other-ai-developer-tools

• What Matters: Attackers are placing fake AI developer tools directly in front of users who are actively looking for them, then relying on copy-and-paste install behavior to execute malicious code locally. That creates immediate access to source code, credentials, and environment data from the developer workstation. This is supply chain risk at the endpoint. Most organizations don’t have strong controls over what developers can download, run, or stage locally, which means these tools become part of the attack surface the moment they’re executed. The gap isn’t awareness. It’s control.
• What’s Overhyped: The AI angle makes this feel new, but the underlying tactic is familiar. Impersonation and malvertising have been working for years. What’s changed is how quickly attackers can align those tactics with whatever developers are already trying to use.

Ghost Campaign Uses 7 npm Packages to Steal Crypto Wallets and Credentials, The Hacker News

https://thehackernews.com/2026/03/ghost-campaign-uses-7-npm-packages-to.html

• What Matters: Malicious NPM packages were able to sit in a trusted ecosystem, get pulled into normal workflows, and then request elevated access to deploy malware. No exploit required. Just trust in the package and the process. This is the compromised workbench playing out in real time. Developers install dependencies, pipelines run as expected, and credentials, wallets, and cloud configs get exposed in the background. Managing dependencies is no longer a hygiene issue. It’s an initial access vector.
• What’s Overhyped: This isn’t specific to NPM. The same risk exists anywhere teams rely on third-party code they don’t fully understand. The real issue is how much implicit trust exists in dependencies, not the ecosystem itself.

Fake interactive Zoom call leads to malicious ScreenConnect download, SC Media

https://www.scworld.com/news/fake-interactive-zoom-call-leads-to-malicious-screenconnect-download

• What Matters: This attack compresses phishing into a real-time interaction. The user joins what looks like a legitimate Zoom call, is prompted to update, and installs remote access software on the spot. There’s no delay between engagement and compromise. The key detail is the simulated environment. It removes hesitation and keeps the user moving forward. Once the software is installed, the attacker has full access under the user’s context.
• What’s Overhyped: It’s easy to frame this as a user awareness problem. In practice, timing and context matter more than training. If users can install unsigned or unapproved software, the system is depending on them to make the right call every time, under time pressure.

Cisco Firewall Vulnerability Exploited as Zero-Day in Interlock Ransomware Attacks, SecurityWeek

https://www.securityweek.com/cisco-firewall-vulnerability-exploited-as-zero-day-in-interlock-ransomware-attacks/

• What Matters: A firewall management flaw was exploited as a zero-day to achieve unauthenticated remote code execution and quickly pivot into ransomware activity. The entry point wasn’t obscure. It was exposed management infrastructure. This highlights how little time exists between disclosure and exploitation. Traditional patch cycles assume a buffer that no longer exists. Once a vulnerability is understood, attackers are already working to turn it into access.
• What’s Overhyped: Focusing only on patch speed misses the larger issue. Systems that centralize access, like firewalls, will continue to be targeted first. Even fast patching doesn’t remove the exposure window.

Threat groups target cyber-physical systems to disrupt critical infrastructure providers, Cybersecurity Dive

https://www.cybersecuritydive.com/news/threat-groups-target-cyber-physical-systems-to-disrupt-critical-infrastruct/815074/

• What Matters: Attackers are gaining access to industrial control systems by abusing exposed interfaces and default credentials, often without needing exploits. In some cases, that access is already translating into operational disruption. These systems weren’t built for internet exposure, but operational demands keep pushing them into accessible environments. Once connected, they inherit the same risks as IT systems without the same level of defensive control.
• What’s Overhyped: The focus often lands on advanced attackers and nation-state capability. In reality, many of these environments are accessible through basic weaknesses. Exposure and default access are still doing most of the work for attackers.