EN PT-BR
← Voltar ao feed
high finance cryptocurrency technology education

Weekly roundup: Trezor implant, LLMJacking up to $1M, Chainalysis breach, Synology ransomware, SAP exploits within days

1 fonte(s) · 1 atualização(ões) · publicado 06/11/2026 11:40 · coletado há 1 hora

SecurityWeek's weekly roundup covers several security stories. Mandiant reports LLMJacking cases that caused fraudulent AI bills of up to $1 million. Trezor is warning buyers of devices sold by the reseller CryptoBilis about an alleged seed-stealing implant. Chainalysis widened its breach disclosure to include password hashes and live tokens. Synology warns that a flaw fixed in December 2024 is now being used in ransomware attacks against outdated DSM versions. This record is based on the roundup alone. No vendor advisories or primary sources were checked.

Entenda a falha

A vulnerabilidade

Two CVEs are listed. CVE-2024-58351 is a code injection flaw in Flowise that lets an attacker run commands on the host. The articles call it old and do not give affected or fixed versions. Era counted 4,372 exposed Flowise instances. Of those whose version it could identify, 703 had no authentication and 46 appeared to share the vulnerable setup from its honeypots. CVE-2026-40217 is a LiteLLM flaw that Mandiant says was used in one intrusion to upgrade ordinary API keys to full admin credentials. The attacker apparently needs a valid ordinary key, and no versions are given. A Synology DSM flaw fixed in December 2024 (Synology-SA-24:20) is also being used in ransomware attacks, but Synology does not name the specific flaw.

Como funciona

The articles give little mechanism. For Flowise, the Greeter framework sent thousands of chat messages containing only the word hello to exposed instances, and the code injection flaw then let it run commands. After that it mapped the host, queried cloud metadata services, searched for SSH keys and credentials, tried to persist and tested running payloads from memory. For LiteLLM, the articles say only that an ordinary key was turned into an admin credential, not at which step the software fails.

Por que é vulnerável

The articles do not state the root cause of either flaw. CVE-2024-58351 is classed as code injection, which implies input is treated as code, but no detail is given. CVE-2026-40217 is described only by its effect, which is that key privileges can be raised to admin. Fix details are not given either. The roundup does note that many exposed Flowise instances run without authentication, which widens the exposure.

Como foi descoberta

Era researchers observed Greeter hitting Flowise honeypots and exploiting CVE-2024-58351, and they detailed the framework. Mandiant found the LiteLLM flaw CVE-2026-40217 in use while investigating LLMJacking cases this year, in which stolen or exposed cloud credentials and API keys were the main way in. Synology disclosed that a flaw it fixed in December 2024 is now exploited in ransomware attacks. This record rests only on a SecurityWeek roundup, and no vendor advisories were checked.

PoC existente, privada

O exploit existe, mas quem o tem não o publicou.

The Greeter framework is a working exploitation toolkit seen in attacks on Flowise honeypots that abuses CVE-2024-58351 for command execution, host mapping, credential search and persistence attempts. The articles do not say it was published, so it counts as private. No PoC is mentioned for CVE-2026-40217, which was only seen exploited in a real intrusion. Because exploitation is already happening, patching Flowise and LiteLLM and rotating exposed keys is urgent.

Explicação gerada por IA a partir da matéria e dos artigos, em 06/11/2026 12:18.

Detalhes técnicos

  • Greeter framework / Flowise: Era researchers detail a Chinese-language exploitation framework that sent thousands of "hello" chat messages to Flowise honeypots. It abused the code injection flaw CVE-2024-58351 to run commands, map the host, query cloud metadata services, search for SSH keys and credentials, attempt persistence and test in-memory payload execution. Era counted 4,372 exposed Flowise instances. Of those whose version it could identify, 703 lacked authentication and 46 appeared to share the vulnerable setup used in its honeypots.
  • LLMJacking: Mandiant saw fraudulent bills above $100,000 per day and up to $1M in total at some organizations. Initial access came mainly from stolen or exposed cloud credentials and API keys. In one intrusion the attackers exploited a LiteLLM flaw (CVE-2026-40217) to upgrade ordinary keys to admin credentials. Google Threat Intelligence Group found hijacked access sold through "LLM Mixers", including the Jet Night token routing service, which offers unlimited plans for about $60 a month.
  • Phishing: Push Security documented fake OpenAI, SpaceX and Anthropic event registration pages. The Microsoft option runs a device code phishing flow through the Kali365 phishing-as-a-service platform. The campaign uses Google and Zapier short links, Cloudflare challenges and domains rotated every few weeks. Push links the operator, with medium-high confidence, to earlier fake Zoom event pages.
  • Trezor: A buyer reported an added circuit board with a cellular module and antenna inside a Trezor Safe 5 bought from CryptoBilis. The board allegedly captures the recovery phrase shown on screen and sends it over a mobile network. Trezor has not verified the report, and no stolen funds are reported. CryptoBilis also sold tampered Ledger devices.
  • Chainalysis: On November 2 the company found that names, usernames, password hashes and live tokens for all its users were exposed. It first said only non-sensitive data was accessed. It has taken several products offline, including KYT, and is contacting KYT customers whose accounts show unauthorized access.
  • SAP: Onapsis and Recorded Future report that the median time from a security note to first verified exploitation or public proof-of-concept was 1–3 days for flaws patched in 2025–2026, against 489 days for 2021. Ransomware and extortion groups in SAP-related content rose from 8 to 26 between 2023 and 2025. SAP CVEs with confirmed exploitation or public proof-of-concept code doubled from 9 to 18. Underground listings for SAP access were priced at $500–$30,000.
  • Synology: Attackers are abusing a flaw fixed in December 2024 (advisory Synology-SA-24:20) against outdated DSM releases to encrypt data and extort victims. The company does not name the specific flaw.
  • Other items: HIBP loaded about 851,000 email addresses from Harvard's November 2025 breach, which was tied to a ShinyHunters extortion attempt. Alexander Gurevich (now Alexander Block) was extradited to the US over the Nomad Bridge exploit in August 2022, worth about $2.89M. Anthropic added security and admin controls to Claude Enterprise. Google donated Agent Substrate and gVisor to the CNCF.

Remediação

  • Synology advises updating DSM immediately and enabling automatic updates.
  • Trezor advises customers who bought from CryptoBilis since March 2026 to move funds to a new wallet with a new seed.
  • Defenders should rotate and protect cloud credentials and API keys, patch Flowise, LiteLLM and SAP promptly, and watch for device code phishing.

Linha do tempo

  1. securityweek.com ↗ · pub. 06/11/2026 11:40 · coletado 06/11 12:00 primeiro registro

    Weekly roundup: Trezor implant, LLMJacking up to $1M, Chainalysis breach, Synology ransomware, SAP exploits within days

    ler a análise desta fonte

    SecurityWeek's weekly roundup covers several security stories. Mandiant reports LLMJacking cases that caused fraudulent AI bills of up to $1 million. Trezor is warning buyers of devices sold by the reseller CryptoBilis about an alleged seed-stealing implant. Chainalysis widened its breach disclosure to include password hashes and live tokens. Synology warns that a flaw fixed in December 2024 is now being used in ransomware attacks against outdated DSM versions. This record is based on the roundup alone. No vendor advisories or primary sources were checked.

    Technical Details

    • Greeter framework / Flowise: Era researchers detail a Chinese-language exploitation framework that sent thousands of "hello" chat messages to Flowise honeypots. It abused the code injection flaw CVE-2024-58351 to run commands, map the host, query cloud metadata services, search for SSH keys and credentials, attempt persistence and test in-memory payload execution. Era counted 4,372 exposed Flowise instances. Of those whose version it could identify, 703 lacked authentication and 46 appeared to share the vulnerable setup used in its honeypots.
    • LLMJacking: Mandiant saw fraudulent bills above $100,000 per day and up to $1M in total at some organizations. Initial access came mainly from stolen or exposed cloud credentials and API keys. In one intrusion the attackers exploited a LiteLLM flaw (CVE-2026-40217) to upgrade ordinary keys to admin credentials. Google Threat Intelligence Group found hijacked access sold through "LLM Mixers", including the Jet Night token routing service, which offers unlimited plans for about $60 a month.
    • Phishing: Push Security documented fake OpenAI, SpaceX and Anthropic event registration pages. The Microsoft option runs a device code phishing flow through the Kali365 phishing-as-a-service platform. The campaign uses Google and Zapier short links, Cloudflare challenges and domains rotated every few weeks. Push links the operator, with medium-high confidence, to earlier fake Zoom event pages.
    • Trezor: A buyer reported an added circuit board with a cellular module and antenna inside a Trezor Safe 5 bought from CryptoBilis. The board allegedly captures the recovery phrase shown on screen and sends it over a mobile network. Trezor has not verified the report, and no stolen funds are reported. CryptoBilis also sold tampered Ledger devices.
    • Chainalysis: On November 2 the company found that names, usernames, password hashes and live tokens for all its users were exposed. It first said only non-sensitive data was accessed. It has taken several products offline, including KYT, and is contacting KYT customers whose accounts show unauthorized access.
    • SAP: Onapsis and Recorded Future report that the median time from a security note to first verified exploitation or public proof-of-concept was 1–3 days for flaws patched in 2025–2026, against 489 days for 2021. Ransomware and extortion groups in SAP-related content rose from 8 to 26 between 2023 and 2025. SAP CVEs with confirmed exploitation or public proof-of-concept code doubled from 9 to 18. Underground listings for SAP access were priced at $500–$30,000.
    • Synology: Attackers are abusing a flaw fixed in December 2024 (advisory Synology-SA-24:20) against outdated DSM releases to encrypt data and extort victims. The company does not name the specific flaw.
    • Other items: HIBP loaded about 851,000 email addresses from Harvard's November 2025 breach, which was tied to a ShinyHunters extortion attempt. Alexander Gurevich (now Alexander Block) was extradited to the US over the Nomad Bridge exploit in August 2022, worth about $2.89M. Anthropic added security and admin controls to Claude Enterprise. Google donated Agent Substrate and gVisor to the CNCF.

    Remediation

    • Synology advises updating DSM immediately and enabling automatic updates.
    • Trezor advises customers who bought from CryptoBilis since March 2026 to move funds to a new wallet with a new seed.
    • Defenders should rotate and protect cloud credentials and API keys, patch Flowise, LiteLLM and SAP promptly, and watch for device code phishing.