ThreatsDay Roundup: AI-Driven Attacks, Exposed Secrets and Zero-Day Exploits
Summary
This week’s cybersecurity roundup covers ATM jackpotting sanctions, blockchain-based malware concealment, 543,699 valid exposed credentials, code execution in Unsloth model inspection, and zero-days used in an attack on DIVD.
Key points
- OFAC sanctioned 10 targets tied to a Tren de Aragua ATM jackpotting scheme that allegedly stole at least $40.73 million.
- Threat actors are using public blockchains to conceal malware instructions, a technique known as EtherHiding.
- A study found 543,699 valid credentials exposed in public GitHub repositories as of July 2026.
- Inspecting a model in Unsloth Studio could trigger Python code execution; the issue was fixed in version 2026.6.9.
- Two Zammad zero-days were chained in an attack on DIVD, enabling session hijacking, remote code execution, privilege escalation, and data exfiltration.
- Researchers described cache key injection that can enable cache poisoning, data leakage, denial of service, and, in specific conditions, stored XSS.
- Google Threat Intelligence Group reported increases in disclosed vulnerabilities and exploitation during 2026, including AI-assisted findings leading to RCE.
Article Details
- Event Type
- Weekly cybersecurity news bulletin covering sanctions related to ATM jackpotting, blockchain-based malware concealment, exposed credentials, vulnerabilities and exploitation, cybercrime prosecutions, and security product developments.
- Impact
- Reported impacts include at least $40.73 million stolen in alleged ATM jackpotting attacks; 543,699 valid credentials exposed in public GitHub repositories; a Zammad compromise that enabled session hijacking, remote code execution, privilege escalation to root, and data exfiltration; and a vulnerability in Unsloth Studio that could execute Python code during model inspection and expose data or credentials. The bulletin also describes a $16 million cryptocurrency fraud case and two individuals sentenced for a phishing and business-email payment-diversion scheme.
CVE
CVE-2025-4632“The incident started with exploitation of a known Samsung MagicINFO flaw [CVE-2025-4632],” Huntress said.CVE-2026-102489Two zero-day vulnerabilities in the open-source Zammad ticketing system, CVE-2026-102489 and CVE-2026-102490, have been chained by threat actors to break into the Dutch Institute for Vulnerability Disclosure (DIVD).CVE-2026-102490Two zero-day vulnerabilities in the open-source Zammad ticketing system, CVE-2026-102489 and CVE-2026-102490, have been chained by threat actors to break into the Dutch Institute for Vulnerability Disclosure (DIVD).
People
Ari RedbordIdentified as Global Head of Policy at TRM Labs; commented on Tren de Aragua's alleged use of ATM malware and cryptocurrency.Chen YixinIdentified as head of China’s Ministry of State Security; discussed AI's implications for security and cyberattack capabilities.Chijioke Timothy OdimegwuSentenced to 189 months in federal prison for his role in an international cyber intrusion scheme involving phishing and payment diversion.Harafat MogajiSentenced to 189 months in federal prison for his role in an international cyber intrusion scheme involving phishing and payment diversion.Trung Nguyen VanVietnamese national charged for an alleged role in a cryptocurrency wire-fraud scheme.Zero SalariumSecurity researcher who outlined a process-injection technique using a console process's stdin pipe and WriteFile().
Threat Actors
Malware
Vendors
CloudflareCloudflare has announced plans to become a public Certificate Authority (CA) that can issue quantum-safe digital certificates that websites need to encrypt traffic and prove their identity to visitors.GlobalSignIn addition, Cloudflare has agreed to acquire established, publicly trusted Root CA key material from GlobalSign to ensure certificates work on older smartphones, operating systems, and devices that no longer receivePillar Security“Selecting a model in the UI caused the backend to download and run Python code shipped inside that model’s HuggingFace repository,” Pillar Security said.Truffle SecurityA new study from Truffle Security has found 543,699 unique credentials exposed in public GitHub repositories that were still valid as of July 2026 despite the platform’s security measures to prevent leakage of such data.
Products
AnyDesk“Attackers then deployed a rogue AnyDesk instance (after three tries), created a new local admin account, and disabled Defender protections.AWS Secrets ManagerWe placed it inside a canary secret in AWS Secrets Manager, where an agent exploring the account could discover it.GitHubIt also highlights active abuse of ATM jackpotting, EtherHiding, Unsloth, Zammad, and exposed GitHub credentials, showing that simple weaknesses still power many modern attacks.Google ChromeEarlier this February, Google said it’s developing an evolution of HTTPS certificates based on Merkle Tree Certificates (MTCs) for use in its Chrome web browser.K3 Swarman internal review after a July 2026 report from Mindgard found that its AI models, Kimi K2.6 and K3 Swarm, could bypass safety guardrails and generate dangerous information, including providing plans forKimi K2.6is conducting an internal review after a July 2026 report from Mindgard found that its AI models, Kimi K2.6 and K3 Swarm, could bypass safety guardrails and generate dangerous information, including providingSamsung MagicINFO“The incident started with exploitation of a known Samsung MagicINFO flaw [CVE-2025-4632],” Huntress said.SignalSignal has introduced on-device encrypted backups for iPhones and iPads with version 8.30, a feature that was already supported on Android, Linux, macOS, and Windows.Unsloth StudioUnsloth Studio could execute Python code just by inspecting a model.ZammadIt also highlights active abuse of ATM jackpotting, EtherHiding, Unsloth, Zammad, and exposed GitHub credentials, showing that simple weaknesses still power many modern attacks.
Tools
Blockchain Dead DropsThis technique, referred to as EtherHiding, is part of a broader approach called Blockchain Dead Drops (BDD).Context BombsIn July 2026, Tracebit detailed a technique called Context Bombs that uses prompt injections as a way to trip an AI model provider’s runtime safety checks and prevent it from taking malicious actions.EtherHidingIt also highlights active abuse of ATM jackpotting, EtherHiding, Unsloth, Zammad, and exposed GitHub credentials, showing that simple weaknesses still power many modern attacks.
Countries
ChinaChen Yixin, head of China’s Ministry of State Security, said AI has implications for political security, institutional security, and ideological security, with “hostile forces” using synthetic content to spreadPakistan@Rayhucker for sample access and sales inquiries, while an OSINT investigation linked the account to a Pakistan-based individual through historical display names and public profiles.United StatesAir Force, the defendants “attacked business victims across the United States with email ‘spamming’ and phishing campaigns to steal usernames and passwords for victims’ employee email accounts,” the Justice Department