Trezor phishing wave hits 347,000 users after Brevo breach as JFrog, GitLab, Sogou flaws surface

Trezor phishing wave follows Brevo email vendor breach

A breach at email marketing vendor Brevo fed a phishing campaign against hardware-wallet maker Trezor's customer base, the company confirmed on September 11, 2026. The attack targeted 347,000 email addresses and 2,500 users clicked an embedded malicious link. The disclosure adds a second supply-chain hit to the wallet vendor in days, on top of a separate shipping-data exposure covering 81,000 customers' home addresses reported the same week. Practitioners handling customer email pipelines should treat vendor-side compromise as a direct phishing vector: the attacker only needs a working mailbox to bypass brand-trust training.

Hardware-wallet attack surface widens with Trezor T exploit claim

Alongside the phishing wave, US cybersecurity firm Unciphered said it hacked a Trezor T hardware wallet by exploiting a hardware vulnerability, a claim tracked across two independent posts on September 12, 2026. The disclosure is the second breach affecting a hardware-wallet company in a short window, reinforcing that device-level flaws, not just phishing, remain in scope. For practitioners keying seed material and signing workflows, the immediate check is whether devices shipped in the affected window require firmware updates or replacement. The shipping breach and phishing wave both widen the set of users whose addresses and habits are now in adversary hands.

Maximum-severity path traversal in GitLab demands immediate patching

GitLab urged administrators to patch on September 11, 2026 against a maximum-severity path traversal vulnerability tracked as CVE-2026-85706. Path traversal flaws let an attacker read or write outside the intended directory tree, and at maximum severity they typically yield code execution or full data exfiltration. Operators running self-managed GitLab instances should confirm the patch is applied before the next code push lands on the server. Unpatched self-hosted source-control servers are a standard staging ground for supply-chain follow-on attacks.

JFrog Artifactory flaw chain granted admin control and backdoors

Attackers chained two flaws in JFrog Artifactory, the repository software build pipelines pull from, to take administrator control of self-hosted servers and plant backdoors, cloud security firm Wiz reported on September 11, 2026. Wiz observed the attacks between August 15 and September 8. JFrog had fixed both flaws before then, so only servers that had not been updated were open to them.

Build-pipeline repositories are high-value targets because a compromised binary flows downstream to every consumer; teams should verify Artifactory patch levels against JFrog's advisories before the next release. Self-hosted artifact servers that lag behind vendor fixes remain a soft entry point for supply-chain compromise.

China-linked group exploited Sogou Input Method for GRAYRABBIT backdoor

A China-linked hacking group tracked as UNC3569 exploited a flaw in Sogou Input Method, one of the most widely used Chinese-character typing tools on Windows, to install a backdoor called GRAYRABBIT, Gen Digital said in research published September 11, 2026. The attack chain started with a crafted link and ended with the attacker able to do anything the logged-in user could do. Tencent, which owns Sogou, did not yet have a public fix in the evidence reviewed. Endpoint teams should watch for Sogou child processes spawning unexpected outbound connections, and treat any update notice from the vendor as urgent given the level of access demonstrated.

AI platforms become an attack surface as state actors weaponize Claude

Anthropic warned on September 11, 2026 that cybercriminals and state-sponsored hackers are using its Claude models for cyberattacks, weapons design, propaganda and mass surveillance between December 2025 and August 2026, in what it labels Generative Threat Groups (GTGs). A separate Anthropic disclosure the same day detailed how a Russian state-sponsored actor branded GTG-20006 used Claude to rebuild malware after detection, aligning with a cluster linked to Midnight Blizzard. Separately, Huntress documented campaigns weaponizing Claude Artifacts, shared AI conversations, sponsored search results and ClickFix-style lures to host malicious content and trick users into installing malware.

The pattern shifts part of the encoding and cryptography threat model from code-level flaws to prompt-level abuse, where the trust boundary is the AI's output rather than the application's input.

Cloudflare CASB ships automatic remediation policies for SaaS risk

Cloudflare introduced automatic remediation policies for Cloudflare CASB on September 11, 2026, building event-driven logic on its developer platform to revoke risky file shares and send webhooks without manual intervention. The release gives security teams a native automation engine inside the CASB product rather than a bolt-on workflow tool. For teams using Cloudflare as a control plane, the change lets them close risky sharing on SaaS files as soon as the CASB detector fires, a workflow that previously required a custom integration. Teams reviewing their SaaS posture can pair the policy engine with Cloudflare's existing access controls and audit exports.