Daily Cyber & AI Briefing with Michael Housch. This episode was published automatically and includes the assembled audio plus full transcript.
Today’s cyber and AI risk landscape is marked by a surge in critical vulnerabilities, active exploitation, and a rapidly evolving regulatory environment. Security leaders are being tested on multiple fronts: from urgent patch management to the governance of shadow AI, and from defending against sophisticated supply chain attacks to anticipating the next wave of AI-driven threats. Let’s break down the key developments shaping today’s risk environment, explore their practical implications, and consider what security leaders need to prioritize right now.
Let’s start with the wave of critical vulnerabilities impacting some of the most widely deployed enterprise security products. Over the past 24 hours, vendors including Cisco, Check Point, Kaspersky, and Tanium have all issued urgent patches for flaws that, in many cases, are already being exploited in the wild. Cisco’s Identity Services Engine, or ISE, is a particular concern, with active attacks targeting environments that haven’t yet applied the latest fix. Check Point, Kaspersky, and Tanium have addressed vulnerabilities that open the door to remote code execution or privilege escalation—attackers’ favorite footholds for lateral movement or data theft.
For organizations running these products, the message is clear: immediate patch deployment is non-negotiable. But it doesn’t stop there. Environments with complex or distributed infrastructure—think hybrid cloud, multiple branch offices, or remote workforces—face additional challenges. Patching at scale requires coordination, testing, and sometimes downtime, but the risk of delay is now demonstrably higher. Attackers are moving quickly to reverse-engineer patches and weaponize exploits, often within hours of disclosure. That means continuous monitoring for exploitation attempts is just as critical as the patching itself. Watch for unusual authentication attempts, privilege escalations, or unexpected outbound traffic—these are often the first signs that a vulnerability is being targeted.
Shifting focus to network infrastructure, MikroTik routers are under active attack, with hackers exploiting vulnerabilities that allow unauthenticated takeover of devices. This is particularly worrisome because MikroTik hardware is popular in both enterprise and service provider environments. Once compromised, these routers can be conscripted into botnets, used as launchpads for lateral movement, or serve as persistent footholds for further compromise. The fact that attackers don’t even need credentials to take control amplifies the urgency. Organizations using MikroTik should prioritize patching, review their network segmentation to limit the blast radius of a potential breach, and monitor for unusual outbound connections that might indicate a device has been hijacked.
Fortinet’s FortiGate appliances are also in the spotlight, with news that a so-called “1-day” vulnerability—meaning a flaw that’s just been patched but is not yet widely remediated—is being sold on underground forums. FortiGate devices are a backbone of security infrastructure for many enterprises, providing firewall and VPN services. The sale of this exploit on criminal marketplaces suggests that widespread attacks could be imminent. Security teams should be monitoring vendor advisories closely, applying patches as soon as they become available, and reviewing firewall and VPN logs for any signs of suspicious activity. This is a classic example of the shrinking window between vulnerability disclosure and exploitation, and it reinforces the need for rapid, automated patch management wherever possible.
Now, let’s talk about the human element—specifically, the rise of shadow AI in the enterprise. A new report finds that 67% of employees are using unapproved AI tools at work. That’s two out of every three workers bypassing official channels to leverage generative AI, coding assistants, or other productivity tools. While the intent may be to boost efficiency, the risks are significant: data leakage, compliance violations, and the introduction of unmonitored, potentially insecure applications into the environment. For CISOs, this is a wake-up call. AI governance policies can’t be theoretical—they need to be rolled out, communicated, and enforced. This means increasing user awareness about the risks of unauthorized AI use, deploying monitoring tools to detect shadow AI activity, and working with business units to provide secure, approved alternatives that meet employees’ needs without compromising the organization’s risk posture.
The supply chain threat landscape is also evolving, with attackers targeting code repositories and exploiting SaaS integrations in new ways. One novel attack, dubbed the “Shai-Hulud Worm,” is hijacking AI coding assistant sessions to compromise SaaS code repositories. This is a significant supply chain risk because compromised code can propagate downstream to customers and partners, multiplying the impact. Security leaders should review their SaaS access controls, enforce multi-factor authentication for critical development tools, and monitor for anomalous activity within code repositories. The lesson here is that the supply chain is only as strong as its weakest link, and attackers are increasingly targeting the tools and integrations that developers rely on every day.
The risks of third-party integrations are further highlighted by a large-scale supply chain attack involving Brevo widgets. Attackers have turned these widgets into a malware delivery channel, impacting over 100,000 websites. This demonstrates the inherent risk of embedding third-party scripts and components into web applications. Organizations should maintain an up-to-date inventory of all third-party scripts, implement content security policies to limit the execution of untrusted code, and be prepared to respond quickly to widespread web-based compromises. The scale of this attack serves as a reminder that even trusted vendors can become vectors for malware delivery if their supply chain is compromised.
Turning to the consumer and enterprise malware front, the MovieReaper malware campaign is spreading through pirated movie torrents and using the Solana blockchain for command-and-control. This is an example of attackers leveraging decentralized technologies to evade takedown efforts. By using the blockchain, they make it much harder for defenders to disrupt their infrastructure. MovieReaper targets both individuals and organizations, potentially leading to ransomware infections or data theft. Security teams should block access to known malicious torrent sources and monitor for unusual network activity related to blockchain transactions. The use of popular blockchains for command-and-control is likely to become more common, so defenders need to adapt their detection strategies accordingly.
As AI adoption accelerates, organizations are scrambling to secure and manage their AI fleets. A growing ecosystem of governance tools is emerging to help with this challenge, but the regulatory environment remains fragmented. Some governments are shelving AI legislation, while others—China, for example—are framing AI as a security threat rather than an existential risk. This divergence creates uncertainty for security executives, who must ensure their governance frameworks are both robust and adaptable. Staying abreast of evolving best practices and regulatory expectations is essential. This means regularly reviewing and updating AI governance policies, engaging with industry groups and regulators, and ensuring that governance tools are flexible enough to accommodate new requirements as they emerge.
On the policy front, the US Cybersecurity and Infrastructure Security Agency, or CISA, has released new guidance on the use of decoys—sometimes called deception technologies—in cyber defense. At the same time, policymakers are debating whether “nuclear-style” safeguards should be applied to AI, reflecting growing concern over the potential for catastrophic misuse. For CISOs, this is a signal to evaluate the applicability of deception technologies within their own environments and to prepare for potential regulatory shifts in AI oversight. Deception can be a valuable tool for detecting and slowing attackers, but it requires careful planning and integration with existing detection and response processes.
A recent analysis from Unit 42 highlights the complex relationship between identity management, cloud harnesses, and supply chain security. The report emphasizes the “uncomfortable space” where vulnerabilities can be exploited—often at the intersection of identity and cloud infrastructure. Security leaders should prioritize identity governance, review cloud access policies, and assess supply chain dependencies for potential weak points. The takeaway is that identity is increasingly the new perimeter, and attackers are adept at finding and exploiting gaps in identity and access management.
Zooming out, the convergence of AI and cyber operations is accelerating. Both attackers and defenders are leveraging AI for automation, reconnaissance, and exploitation. For defenders, this means investing in AI-enabled security tools that can keep pace with adversarial capabilities. It also requires ongoing staff training to ensure that teams can effectively use these tools and adapt to new tactics. The bar for defensive innovation is rising, and organizations that fail to keep up risk being outmaneuvered by more agile, AI-powered adversaries.
Let’s take a step back and consider the strategic implications of these developments. First, patch management and vulnerability response must be accelerated, especially for widely deployed security infrastructure. The window