In Other News: Trezor Implant, LLMJacking Bills Hit $1 million, SAP Flaws
Hardware seed implants, 1 million dollar LLMJacking bills, and SAP exploits landing within three days highlighted widespread enterprise infrastructure threats.
By Dillip Chowdary โข Nov 06, 2026 โข Source: SecurityWeek
Hardware security alerts, artificial intelligence infrastructure exploits, enterprise software weaponization, and expanded data theft disclosures surfaced across multiple independent investigations summarized in SecurityWeek's report. Trezor alerted cryptocurrency users after a customer discovered a rogue cellular implant inside a Trezor Safe 5 purchased through former Southeast Asian reseller CryptoBilis. Meanwhile, cloud security researchers tracked automated LLMJacking attacks that generated over 100,000 dollars in fraudulent model inference charges in a single day, while Onapsis and Recorded Future documented enterprise SAP vulnerabilities being actively exploited in the wild within one to three days of disclosure.
This security update examines the concurrent waves of hardware tampering, AI API hijacking, enterprise ransomware campaigns, and credential disclosures threatening technical teams. The coverage details tactical risks for security engineers running LiteLLM and Flowise agent frameworks, IT administrators managing unpatched Synology NAS hardware or SAP installations, and organizations assessing third-party breach disclosures from Chainalysis and Harvard University.
What broke in In Other News
Hardware tampering emerged as a direct physical threat after an unauthorized secondary circuit board fitted with a cellular module and antenna was found embedded inside a Trezor Safe 5 hardware wallet. The hardware was sold through former reseller CryptoBilis, which had also distributed tampered Ledger devices. Software defenses failed across multiple developer platforms, highlighted by Chinese-language exploitation framework Greeter attacking Flowise instances by abusing CVE-2024-58351 with repeated chat messages containing the word hello. In cloud AI environments, attackers weaponized a LiteLLM vulnerability tracked as CVE-2026-40217 to escalate standard API keys into administrative credentials, enabling illicit model queries at scale.
Data exposure expanded significantly at Chainalysis, which uncovered on November 2 that full user records containing names, usernames, password hashes, and active session tokens had been compromised, reversing an earlier assessment that only non-sensitive records were accessed. Synology confirmed that an unpatched vulnerability originally resolved in its December 2024 advisory Synology-SA-24:20 is undergoing active exploitation in the wild by ransomware operators seeking to encrypt network storage systems. Meanwhile, Have I Been Pwned indexed roughly 851,000 unique email addresses tied to a November 2025 breach of Harvard University fundraising databases claimed by extortion collective ShinyHunters, surfacing names, donor amounts, birth locations, and ethnic backgrounds.
Who is exposed by In Other News

Cryptocurrency holders who purchased Trezor devices from CryptoBilis since March 2026 face physical exposure from rogue cellular boards transmitting seed phrases during setup. Enterprise teams hosting Flowise workflow tools are widely targeted, with security firm Era identifying 4,372 exposed instances online, including 703 configurations operating completely without authentication and 46 matching vulnerable honeypot signatures. Organizations utilizing cloud-hosted artificial intelligence models and LiteLLM routers are exposed to automated LLMJacking, which Mandiant documented racking up fraudulent bills of 100,000 dollars in 24 hours and reaching 1 million dollars in aggregate damages across compromised enterprises.
Enterprises relying on Chainalysis compliance tools face credential exposure and operational disruption, particularly customers of the KYT transaction monitoring product that was pulled offline after systems flagged unauthorized access patterns. Global businesses running enterprise resource planning platforms face accelerated compromise, with threat intelligence showing SAP vulnerabilities patched in 2025 and 2026 being weaponized in one to three days, while underground criminal forums list unauthorized SAP corporate network access points priced between 500 dollars and 30,000 dollars. In addition, corporate marketing personnel remain exposed to fake event registration sites mimicking OpenAI, SpaceX, and Anthropic that deploy Kali365 device code phishing flows.
Advertisement
Tech Pulse Daily
Get tomorrow's pulse first
Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.
What to do now about In Other News
Trezor instructs every customer who bought hardware from CryptoBilis since March 2026 to immediately generate a fresh recovery phrase on an untampered device and migrate all digital assets off the existing wallet. Synology administrators must apply current DSM firmware builds immediately across all NAS hardware and turn on automated operating system updates to block active extortion attacks tied to the Synology-SA-24:20 advisory. Flowise operators should place instances behind strict authentication layers, close public network exposures, and apply patches addressing code injection flaw CVE-2024-58351 to prevent memory-based shell execution.
Enterprises deploying machine learning APIs must rotate existing keys, patch LiteLLM against CVE-2026-40217, and enforce strict per-key spending limits to contain LLMJacking before billing cycles spiral. Cloud platform owners can evaluate Google's newly open-sourced Agent Substrate and gVisor projects contributed to the CNCF Sandbox, which isolate agent containers in user space and preserve state in 500 milliseconds. Anthropic enterprise users should activate inference hooks to inspect tool calls via data loss prevention tools, use the centralized management console for SCIM and single sign-on, and review beta security center recommendations against baseline chat and Claude Code configurations.
How the In Other News issue works
The hardware attack against the Trezor Safe 5 functions by physically intercepting the cryptographic seed phrase displayed on the screen using an implanted secondary board, which then transmits the captured phrase across cellular networks using an onboard antenna whenever the unit receives power. In the Flowise attacks, the Greeter framework sends floods of chat queries consisting solely of the greeting hello to trigger code execution through CVE-2024-58351, allowing scripts to probe cloud instance metadata, locate SSH credentials, and test memory payload execution. LLMJacking operators monetize stolen cloud credentials and escalated LiteLLM keys by funneling access into illicit LLM Mixers like the Jet Night routing service, which resells unlimited access to frontier models for approximately 60 dollars monthly.
Threat actors targeting SAP infrastructure exploit an operational window that dropped from a 489-day median in 2021 down to between one and three days following security note publication, according to data from Onapsis and Recorded Future. Attackers rapidly convert public proof-of-concept code into functional exploits, while 26 ransomware and extortion groups now actively target SAP weaknesses. The Push Security phishing campaign functions by offering promotional incentives such as three months of ChatGPT Pro or Grok Heavy, routing marketing personnel through Zapier and Google link shorteners into Kali365 device code authorization flows hosted behind changing Cloudflare-shielded domains.
What is still unknown about In Other News
Trezor has noted that it has not independently verified the hardware implant claims reported by the customer, and forensic investigators have not confirmed whether any cryptocurrency has actually been stolen from the modified Safe 5 unit. Synology has similarly withheld the exact Common Vulnerabilities and Exposures identifier tied to current NAS ransomware activity beyond citing the overarching Synology-SA-24:20 bulletin, leaving the precise software subcomponent under active assault unspecified. Law enforcement authorities have yet to identify the underlying corporate victims or the full operator network behind the Jet Night token mixer documented by Google Threat Intelligence Group.
The full identity and physical location of the threat actor operating the Kali365 phishing campaign also remain unresolved, though Push Security assesses with medium-high confidence that the operator previously managed fraudulent Zoom meeting registration portals. At Chainalysis, the technical intrusion pathway that allowed attackers to obtain password hashes and live tokens alongside customer contact details has not been publicly detailed. In federal court, criminal proceedings continue against Alexander Gurevich, also known as Alexander Block, following his October 29 extradition from Israel to face charges of laundering 2.89 million dollars in WBTC and WETH stolen during the August 2022 Nomad Bridge smart contract compromise.
Developer Action Items
- โ Inventory whether Framework / LiteLLM runs in prod, CI, staging, or on laptops before you debate severity.
- โ Pull the vendor advisory for CVE-2024-58351, CVE-2026-40217 and patch from that page โ not from a social recap.
- โ If you cannot patch today, isolate the service, rotate tokens that sat on the affected surface, and raise the logging floor.
- โ Record the decision and residual risk so the next on-call does not re-litigate whether you are exposed.
In Other News FAQ
What happened to the Trezor Safe 5 hardware wallet?
A buyer found an unauthorized secondary circuit board with a cellular module and antenna inside a Trezor Safe 5 purchased from former reseller CryptoBilis, designed to transmit seed phrases over mobile networks.
How much financial damage does LLMJacking cause?
Mandiant reported that stolen credentials and compromised LiteLLM instances allow attackers to generate fraudulent AI model billing exceeding 100,000 dollars in a single day and up to 1 million dollars in total enterprise damages.
How quickly are SAP vulnerabilities being targeted by attackers?
Research from Onapsis and Recorded Future reveals that the median time between an SAP security note release and active exploitation or public proof-of-concept code dropped to one to three days for 2025 and 2026 flaws.
What user data was exposed in the expanded Chainalysis breach disclosure?
Chainalysis confirmed that names, usernames, password hashes, and active session tokens for all users were accessed, forcing the temporary shutdown of its KYT transaction monitoring product.
Sources
Author
Dillip Chowdary
Writes Tech Bytes coverage of AI, engineering, and the tools that actually ship. Editor of Tech Pulse Daily.
Related on Tech Bytes
Advertisement