In the rapidly shifting landscape of global digital threats, staying ahead of emerging tactics requires a keen eye on the granular developments that often foreshadow broader systemic risks. SecurityWeek\u2019s weekly roundup serves as a vital compass for security professionals, distilling complex vulnerability disclosures, sophisticated supply chain campaigns, and evolving policy directives into a coherent narrative of the current threat environment. This week\u2019s insights underscore the increasing weaponization of AI, the persistent dangers of supply chain interdependencies, and the tightening grip of law enforcement on cybercriminal infrastructure.<\/p>\n
The intersection of artificial intelligence and malicious activity has reached a new level of sophistication with the emergence of PoeLLM. According to detailed research from Black Lotus Labs, this malware has been active since at least April 2026, specifically designed to exploit exposed AI services and open-source infrastructure. The malware primarily targets platforms such as LiteLLM, Ollama, Gotenberg, and Gitea, using them as beachheads for unauthorized cryptocurrency mining and the expansion of a growing botnet.<\/p>\n
What sets PoeLLM apart is its unconventional command-and-control (C&C) architecture. Rather than relying on static IP addresses or hardcoded domains\u2014which are easily flagged by network security tools\u2014the malware performs a clever obfuscation technique: it fetches a poem hosted on a public GitHub repository. By extracting four specific keywords from this poem and converting them into an IP address, the operators can dynamically redirect the infected botnet simply by updating the GitHub content. This method of "Canto Incognito" has proven resilient, with the operator, who is believed to be an Italian-speaking individual, updating the poem 11 times to shift C&C infrastructure. This highlights a growing trend where threat actors leverage legitimate, high-reputation services to mask their illicit operations.<\/p>\n
Supply chain security remains a critical vulnerability, as evidenced by the resurgence of the GhostAction campaign. GitGuardian reports that this malicious operation, which utilizes a secret-stealing GitHub Actions workflow, has successfully infiltrated 772 additional public repositories between August 31 and September 30. By embedding themselves into these workflows, the attackers have targeted a staggering 2,577 sensitive secrets, including SSH keys, Azure and AWS credentials, and database access tokens.<\/p>\n
The campaign has demonstrated remarkable longevity, with data suggesting it never truly ceased, but rather evolved. While the attackers introduced a new exfiltration server to refresh their infrastructure, the underlying playbook mirrors the tactics observed in 2025. This persistence serves as a stark reminder that once a supply chain is compromised, the remediation efforts must be comprehensive. For the 373 users and organizations impacted, the fallout involves not only the immediate loss of credentials but also the daunting task of rotating secrets that may have been exposed for an extended period.<\/p>\n
The long arm of the law continues to close in on those who exploit decentralized finance (DeFi) for illicit gain. A jury has recently convicted 36-year-old Jonathan Spalletta of Maryland for his role in two separate hacks of the Uranium Finance decentralized exchange in 2021. Spalletta, who exploited vulnerabilities within the platform’s smart contracts, successfully siphoned approximately $1.4 million in the first incident, followed by a devastating $53.3 million heist that ultimately forced the exchange to cease operations.<\/p>\n
The subsequent investigation revealed a complex trail of money laundering, with Spalletta utilizing privacy-focused platforms like Tornado Cash to obfuscate the origin of the stolen assets. The proceeds were used to fund a lavish lifestyle, including the purchase of rare collectibles such as antique Roman coins and high-value Magic: The Gathering and Pok\u00e9mon cards. With the fraud and money laundering charges carrying potential sentences of 10 and 20 years respectively, this case stands as a significant milestone in the effort to bring accountability to the volatile and often opaque world of cryptocurrency hacks.<\/p>\n
On the policy front, the Cybersecurity and Infrastructure Security Agency (CISA) is refining its internal retention strategies. Recognizing the difficulty of maintaining a high-caliber cybersecurity workforce, CISA has opted to extend its Cybersecurity Retention Incentive program through fiscal 2027. However, the program is being tightened to address concerns raised by the DHS inspector general regarding previous mismanagement and overly broad application. Under the new criteria, employees must maintain an "exceeds expectations" performance rating and dedicate at least 51% of their time to cyber-specific duties within designated job series. This shift ensures that premium pay is focused on those directly contributing to the agency’s mission-critical security goals.<\/p>\n
Simultaneously, the Operational Technology Cybersecurity Coalition (OTCC) is advocating for a more robust approach to securing the nation\u2019s physical infrastructure. The coalition has proposed a new CISA Binding Operational Directive (BOD) tailored specifically for operational technology (OT) at federal civilian agencies. These agencies manage over 8,000 facilities, ranging from HVAC systems to critical power management and building automation. The OTCC proposal calls for the designation of a senior official accountable for OT security and insists that agencies prioritize CISA\u2019s Cybersecurity Performance Goals. By formalizing these requirements, the coalition aims to close the security gap between traditional IT systems and the often-vulnerable industrial control systems that underpin federal operations.<\/p>\n
While large-scale infrastructure remains a target, everyday consumers and corporate employees continue to face risks from credential stuffing. Domino\u2019s recently confirmed that a segment of its customer base was affected by unauthorized access, with attackers using pairs of email addresses and passwords harvested from previous, unrelated third-party data breaches. While the company stated that its internal systems remained secure and no payment information was stored, the incident forced a mandatory account reset for affected users, highlighting the continued danger of password reuse.<\/p>\n
The geopolitical dimension of cyber threats has also intensified, particularly in South Korea, where the government is probing a series of bank cyberattacks. President Lee Jae Myung has publicly noted the involvement of artificial intelligence in these intrusions, which led to the compromise of sensitive customer personal information. Security analysts at CrowdStrike, who have been monitoring the threat actor’s infrastructure, discovered traces of Claude Code session histories and ARTEX configuration files. Investigators believe with moderate confidence that a Chinese-speaking, financially motivated threat actor is behind these breaches, marking a sophisticated evolution in how AI-integrated tools are being utilized to facilitate financial crime.<\/p>\n
The dark web marketplace ecosystem suffered a significant blow this week with the sentencing of 30-year-old Raheim Hamilton. As a co-creator of Empire Market, which facilitated over four million transactions worth more than $430 million between 2018 and 2020, Hamilton received a 40-year prison sentence and a $5 million fine. The marketplace was a hub for the distribution of illicit narcotics, stolen credentials, and hacking tools. This sentencing serves as a definitive warning to those operating criminal platforms under the shroud of anonymity.<\/p>\n
In the realm of hardware and cloud security, researchers have disclosed CVE-2026-47483, a high-severity vulnerability within Nvidia\u2019s DCGM Exporter. This tool, used for monitoring GPU telemetry, was found to be susceptible to unauthenticated resource exhaustion attacks. By flooding the exporter’s profiling endpoints with requests, an attacker could crash the service and potentially degrade the performance of AI workloads on the host. Scans conducted earlier this year revealed approximately 2,100 internet-exposed hosts leaking telemetry from over 12,000 GPUs. Nvidia has since released a patch in version 4.8.2, and organizations are urged to prioritize the update to prevent potential disruption to their AI infrastructure.<\/p>\n
Finally, the fragility of the open-source ecosystem was underscored by the compromise of version 0.5.144 of the In the rapidly shifting landscape of global digital threats, staying ahead of emerging tactics requires…<\/p>\n","protected":false},"author":20,"featured_media":1075,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[432],"tags":[1216,434,143,1215,435,568,433,902,5,1214],"class_list":["post-1076","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-hardware-security","tag-accountability","tag-chips","tag-cybersecurity","tag-dark","tag-devices","tag-driven","tag-hardware","tag-malware","tag-security","tag-weekly"],"_links":{"self":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/posts\/1076","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/users\/20"}],"replies":[{"embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1076"}],"version-history":[{"count":0,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/posts\/1076\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/media\/1075"}],"wp:attachment":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1076"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1076"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1076"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}tensorlake<\/code> npm SDK. Described by researchers as a "Shai-Hulud" style supply chain attack, the compromised package was engineered to execute a credential-stealing worm upon installation. The malware was capable of harvesting a wide array of sensitive information, including Kubernetes, AWS, and GitHub credentials, and possessed the ability to propagate itself by injecting malicious code into other packages published by the victim. As these threats continue to evolve, the necessity for rigorous package auditing and secure development practices remains paramount for the software engineering community.<\/p>\n","protected":false},"excerpt":{"rendered":"