Axios npm Hack, FortiClient EMS Bugs Exploited
▼ Summary
– Financial industry groups warn that affordable AI deepfakes are enabling widespread identity attacks and call for government action.
– Multiple software supply chain attacks, including on the Axios npm library, have been linked to state actors and triggered widespread data theft.
– Several critical zero-day vulnerabilities were actively exploited in products from companies like Fortinet, Cisco, and Google Chrome.
– New phishing toolkits and techniques, such as EvilTokens for device code attacks, are increasing threats to platforms like Microsoft 365.
– Security leaders emphasize the need to communicate technical risks in business terms and integrate security to reduce operational friction.
The financial sector is mobilizing against a rising tide of AI-driven identity attacks. A coalition of major banking groups has published a joint paper detailing how generative AI tools have made deepfake production cheap and routine for criminals and state actors. The document outlines the scale of the threat and urges policymakers at federal and state levels to implement stronger countermeasures.
In enterprise security, Mimecast has introduced an API-based email security solution designed for rapid deployment. The company’s Chief Product Officer explained that this approach delivers protection equivalent to a traditional gateway without requiring changes to infrastructure, a critical advantage for teams combating business email compromise and credential phishing.
A recent CISO interview highlighted the business imperative of security, noting that mature programs reduce friction in sales and mergers while building organizational trust over time. This focus on driving business value is becoming a core metric for security leadership.
The software supply chain faced significant attacks last week. An unknown attacker compromised the accounts of the main developer behind Axios, a ubiquitous HTTP client library, and published backdoored npm packages. The malicious dependency triggered the installation of droppers and remote access trojans. Google researchers later linked this Axios npm supply chain attack to North Korean hackers, warning that hundreds of thousands of stolen secrets may now be circulating. This incident is connected to a broader wave of intrusions involving the Trivy, KICS, LiteLLM, and Telnyx ecosystems.
Separately, European authorities confirmed that the Trivy supply chain attack enabled the ShinyHunters group to breach European Commission cloud infrastructure, leading to the theft and leak of roughly 340 GB of data.
Critical vulnerabilities in enterprise software were actively exploited. Attackers targeted FortiClient Endpoint Management Server (EMS) using two flaws: a previously patched SQL injection and a new API authentication bypass exploited as a zero-day, tracked as CVE-2026-35616. Emergency hotfixes are available. Cisco also addressed ten vulnerabilities in its Integrated Management Controller, the most severe of which, CVE-2026-20093, allows unauthenticated remote attackers to bypass authentication and gain admin access.
Phishing campaigns are evolving. Researchers report a surge in device code phishing targeting Microsoft 365 users, fueled by the availability of EvilTokens, a specialized phishing toolkit offered as-a-service on Telegram. Google moved quickly to patch a Chrome zero-day, CVE-2026-5281, which was being exploited in the wild. Meanwhile, a source code leak for Anthropic’s Claude Code tool was exploited to spread malware disguised as unlocked software versions.
Broader industry discussions are shifting focus. One analysis argues that the hardest part of a security leader’s job is not identifying risk but compelling action, moving beyond mere threat identification. Another perspective challenges the notion that humans are the weakest link, suggesting the industry’s real problem is poor communication from practitioners to the people they aim to protect.
Video interviews offered practical advice for security leaders on translating technical risk for executives and explained why treating identity management and fraud detection as separate systems creates dangerous gaps.
For consumers, research indicates that government cybersecurity guidance largely focuses on prevention, leaving households with limited support after a smart home device breach. On the technical front, the open-source spam filter Rspamd 4.0.0 was released with significant infrastructure changes that require a migration step before upgrading. Apple added a new security warning in macOS to combat ClickFix attacks, which trick users into running malicious Terminal commands.
In other incidents, US prosecutors charged a man in connection with two hacks of the Uranium Finance crypto exchange that stole over $50 million, funds allegedly spent on trading cards. Toy maker Hasbro confirmed a cyberattack detected on March 28, taking systems offline with recovery expected to take weeks. A TrueConf zero-day vulnerability, CVE-2026-3502, was exploited by suspected China-nexus actors to target government networks in Southeast Asia. Apple extended security updates to more devices in response to the DarkSword exploit kit.
The rise of autonomous AI agents prompted several developments. Microsoft released an open-source Agent Governance Toolkit to provide oversight for agents that can execute tasks without human intervention. Concurrently, researchers introduced SandboxEscapeBench, a benchmark to test if AI agents with shell access can break out of their container sandboxes. Amazon announced AI agents for on-demand penetration testing and DevOps through AWS.
New tools aim to streamline security work. ShipSec Studio is an open-source platform designed to replace ad-hoc scripts with dedicated workflow orchestration for security operations. The SystemRescue 13 recovery environment updated its kernel and added new repair tools. Google introduced location privacy features in Android 17 Beta 3, including one-time access permissions.
March’s notable open-source tools were highlighted, and Intel published a centralized repository of data center performance knowledge on GitHub. Google also rolled out mandatory developer verification for all Android developers to curb malicious apps and made ransomware detection and file restoration generally available for Google Drive.
On the Windows front, a new Insider build features a rebuilt console engine with regex search and significant speed improvements. A concerning study found that malware detectors trained on one dataset often fail when presented with different, real-world samples, highlighting an evasion risk. Microsoft also added High Volume Email capabilities to Exchange Online for automated bulk messaging.
Innovative research explored using existing 5G-Advanced base stations for urban drone detection as a cost-effective alternative to radar. An analysis of Android messaging apps compared the permissions and data access of Messenger, Signal, and Telegram.
Finally, the latest SANS Identity Threats & Defenses Survey reveals that 55% of organizations experienced an identity-related compromise last year, with 26% citing MFA fatigue as a contributing factor. The week closed with a roundup of new cybersecurity products and available job listings.
(Source: Help Net Security)
