Trezor Devices Tampered, SAP Vulnerabilities Exploited Rapidly

Key Cybersecurity Developments This Week

This week saw critical advancements and alarming incidents in the realm of cybersecurity, from hardware wallet security breaches to rushed vulnerability exploitations. Noteworthy updates include the emergence of LLMJacking threats, swift exploitation of SAP flaws, and the alert regarding compromised crypto wallets. Below is a summary of key cybersecurity news that underscores the evolving risk landscape.

Anthropic Enhances Security for Claude Enterprise

Anthropic has rolled out new security features for its AI product, Claude Enterprise. The updated capabilities include a security center, which is currently in beta testing. This center helps organizations validate their configurations against recommended security settings. Additionally, the release provides inference hooks that connect prompts and tool calls to existing data loss prevention systems, allowing for enhanced security checks.

The security center and a centralized console, both currently in beta, streamline the management of single sign-on, SCIM, and other access settings across Claude deployments. This initiative marks an important step toward fortifying AI systems against potential threats.

Phishing Campaign Targets High-Profile Events

A new phishing campaign has surfaced, spoofing registration pages for events related to OpenAI, SpaceX, and Anthropic. Cybercriminals lure victims with enticing rewards aimed at marketing professionals, such as complimentary access to ChatGPT Pro or Grok Heavy. The phishing scheme utilizes a device code phishing method linked to the Kali365 phishing-as-a-service platform, signifying a dangerous trend for sensitive corporate accounts.

This campaign employs dynamic Google and Zapier short links and frequently rotates domains to evade detection, showcasing the sophistication of today’s phishing techniques.

Chainalysis Breach Expands Scope of Compromised Data

On November 2, Chainalysis disclosed a significant breach, revealing that names, usernames, and password hashes for all users had been compromised. Previously, the company had indicated that only non-sensitive information had been accessed. Following the discovery of abnormal activity, Chainalysis has taken several of its products offline, including its KYT transaction monitoring service.

In response to the breach, the company has commenced outreach efforts to impacted customers, particularly those with accounts showing signs of unauthorized access. The gravity of this incident emphasizes the risks associated with centralized data handling.

Google Opens AI Infrastructure Projects to Community

Google has announced the donation of two critical projects—Agent Substrate and gVisor—to the Cloud Native Computing Foundation (CNCF). Agent Substrate serves as an orchestration layer for managing numerous isolated AI agent sessions and can restore sessions in under half a second. gVisor acts as an open-source application kernel that enables untrusted AI-generated code to run securely without direct interaction with the host kernel.

This move fosters a communal and vendor-neutral foundation for AI agent infrastructure that works across Kubernetes clusters, ensuring better security and orchestration capabilities for current AI applications.

Chinese Exploitation Framework Targets Flowise

Researchers from Era have identified a new Chinese-language exploitation framework named Greeter, which has been actively targeting Flowise honeypots. This framework utilizes a simple "hello" message to execute commands by exploiting an uncovered code injection flaw (CVE-2024-58351). The Greeter framework is reported to have interacted with more than 4,300 Flowise instances online, showcasing a need for stronger defenses against such rudimentary but effective attacks.

Evidence suggests that the framework not only attempts to navigate the host system and query metadata services but also targets SSH keys and credentials to establish persistence in the compromised environment.

LLMJacking Cases Lead to Substantial Financial Losses

Investigations by Mandiant shed light on alarming instances of LLMJacking where malicious actors hijacked enterprise access to AI models. Some organizations have lost over $1 million as a result of these intrusions. The attackers primarily gain entry through stolen cloud credentials or exposed API keys, with documented cases showing that companies have suffered significant financial impacts in just one day.

In at least one incident, hackers exploited a LiteLLM vulnerability (CVE-2026-40217) to escalate ordinary user keys to full administrative privileges, facilitating even more severe breaches.

Trezor Users Warned of Tampered Hardware

Trezor has alerted customers who acquired hardware wallets from the reseller CryptoBilis since March 2026 to move their funds to a new wallet. This warning follows a report from a user who discovered an implant embedded within their Trezor Safe 5 wallet. This altered device allegedly contains a circuit board equipped with a cellular module that captures and relays the recovery phrase presented on the device’s screen over mobile networks whenever the device connects.

While no funds have been reported stolen at this time, the situation raises serious concerns about hardware wallet security and the potential for future exploits.

Exploitation of SAP Vulnerabilities Increases

A recent study by Onapsis and Recorded Future reveals a disturbingly quick timeframe for exploiting SAP vulnerabilities. For updates delivered in 2025 and 2026, attackers are reportedly able to exploit flaws within just one to three days of a security note being released. This is a drastic reduction from the median 489 days for vulnerabilities patched in 2021.

The escalation of ransomware and extortion groups targeting SAP-related flaws has increased from eight groups in 2023 to 26 in 2025. Furthermore, reported SAP CVEs with confirmed exploitation have doubled from nine to eighteen within the same timeframe. This trend indicates a growing interest in targeting these widely used enterprise systems.

Ransomware Targets Synology Devices

Synology has issued a warning that attackers are exploiting a previously patched vulnerability from December 2024 in order to launch ransomware attacks. Organizations still using outdated DSM releases are particularly vulnerable, leading Synology to recommend immediate updates and the activation of automatic updates for their systems.

The efforts to exploit this specific flaw underline the need for organizations to maintain current security updates and proactive measures against known vulnerabilities.

Alexander Gurevich in US Court Over Nomad Bridge Exploit

Alexander Gurevich, an Israeli national, faced a US court on October 30 following his extradition from Israel. He is accused of exploiting the Nomad Bridge smart contract in August 2022, resulting in the theft of approximately $2.89 million in virtual currency. Gurevich allegedly laundered the stolen funds by converting them into other cryptocurrencies and dispersing them across multiple wallets to hide their origins. Facing serious charges, he could potentially face up to 20 years in prison if convicted.

Key Takeaways

• Anthropic has enhanced security capabilities in Claude Enterprise, focusing on user management and protection.
• Chainalysis reported a significant breach revealing usernames, password hashes, and access tokens.
• LLMJacking incidents resulted in financial losses exceeding $1 million for some enterprises.
• Trezor customers have been advised to transfer funds after a suspected tampered wallet was discovered.
• The exploitation window for SAP vulnerabilities has significantly decreased, intensifying the threat landscape.

Conclusion

This week’s developments highlight ongoing and emerging threats in the cybersecurity landscape. As organizations adapt to the complex threat environment, proactive measures and vigilance are increasingly crucial to safeguard sensitive information and assets.