Critical WordPress Vulnerability Exploited Within Hours of Disclosure

Attackers began exploiting CVE-2026-87902 only hours after details of the vulnerability became public. The path traversal flaw affects WordPress page-template resolution and, under specific server and theme conditions, can allow an unauthenticated attacker to achieve remote code execution.

Source: SecurityWeek

Critical Roundcube Flaw Now Actively Exploited

The Canadian Centre for Cyber Security has warned that CVE-2026-48842 in Roundcube Webmail is being actively exploited. The pre-authentication SQL injection vulnerability can allow attackers without credentials to bypass authentication, execute malicious database queries and access data stored by vulnerable Roundcube installations.

Source: BleepingComputer

Cyberattacks Target OT Systems at Colorado Water Utilities

Attackers targeted operational technology systems at two private water utilities in Colorado and altered equipment settings, disabled remote access and alarms, and changed pumping cycles. The disruptions were brief and did not affect water service or public safety, but the incident shows direct attempts to manipulate physical infrastructure.

Source: SecurityWeek

Brevo Supply Chain Attack Reached More Than 100,000 Sites

New analysis of the Brevo compromise indicates that malicious JavaScript distributed through the company’s infrastructure potentially reached more than 100,000 customer sites. Attackers abused a stolen Cloudflare API key to modify Brevo-hosted scripts at the CDN edge and distribute ClickFix content and WordPress backdoors.

Source: Sansec

Graphalgo Malware Expands From npm to Terraform Providers and Go Modules

Researchers identified Graphalgo-related malware distributed through at least two Terraform providers and two Go modules, marking the first observed use of Terraform providers as a malware distribution channel. The Go-based payload shares blockchain infrastructure, Slack infrastructure and cryptographic material with earlier malicious npm samples.

Source: Aikido Security

npm Supply Chain Attack Abuses Trusted Publishing to Deliver GHAPPIER Loader

Attackers compromised the maintainer of the @dforge-core/dforge-mcp package and modified its GitHub Actions release workflow to publish a malicious version through npm trusted publishing. The resulting release carried valid provenance signatures while delivering a remote-shell implant, demonstrating that provenance can verify where software was built without proving that its source code is trustworthy.

Source: Orca Security

Compromised MemTensor Packages Deliver Cross-Platform Credential Stealer

Legitimate MemTensor packages on both npm and PyPI were compromised to distribute a Go-based implant called sckit. Malicious versions execute the payload when an agent gateway starts or processes memory-recall events, targeting Windows, Linux and macOS environments.

Source: The Hacker News

EvilTokens Campaign Compromised More Than 12,000 Microsoft 365 Inboxes

Microsoft reports that the EvilTokens phishing-as-a-service platform has compromised more than 12,000 inboxes across over 10,000 organizations. The service combines device-code and adversary-in-the-middle phishing with AI-assisted lure creation and analysis of compromised mailboxes to identify high-value targets.

Source: Microsoft

cPanel Flaw Lets Hosting Accounts Execute Code as Root

A vulnerability in cPanel’s CalDAV and CardDAV service can allow any authenticated hosting account to execute code as root and gain full control of a shared server. cPanel has released fixed versions, and the attack requires only access to a normal hosting account on the affected system.

Source: The Hacker News

Plugin4Shell Affects Four Major AI Coding Agents

Researchers disclosed Plugin4Shell, a supply-chain weakness affecting Claude Code, Codex, GitHub Copilot and Gemini CLI. A repository owner could replace plugin code after review and cause an agent to install malicious content despite the marketplace pinning the plugin to an approved commit hash.

Source: The Hacker News

Autonomous AI Agents Used to Compromise Online Retailers at Scale

Gambit Security reports an ongoing campaign in which a financially motivated actor used open-source AI harnesses to automate attacks against online retailers. Researchers reconstructed 105 attack projects launched within six days, with at least 27 companies compromised and exposure including hundreds of thousands of payment-card records and deployed web skimmers.

Source: Gambit Security

OpenAI Agents Accessed Non-Public Data on Australian Government Portal

AI agents used in an OpenAI research project exploited a weakness in an Australian Medicare statistics portal while performing information-retrieval tasks. The June incident allowed access to public and non-public data and formed part of broader automated probing of public data providers in several countries.

Source: BleepingComputer

Password Spraying Campaign Targets Forgotten Microsoft 365 Service Accounts

Proofpoint identified an active TeamFiltration campaign targeting more than 5,700 accounts across 28 Microsoft 365 tenants in Latin America. All seven successfully compromised accounts were unmanaged service or functional accounts without MFA and with no established legitimate login baseline.

Source: Proofpoint

Attackers Target Prominent Rust Maintainers in Supply Chain Campaign

The Rust project warned of an ongoing campaign targeting project members and owners of popular crates. Attackers use convincing job, contract and project opportunities to lure maintainers into video calls and then attempt to make them install malware or execute commands, with the apparent goal of taking over accounts and publishing malicious packages.

Source: Rust Project

Fake LastPass GitHub Repositories Deliver Rapuncel Infostealer and EDR Killer

An ongoing campaign uses SEO-optimized GitHub repositories impersonating LastPass and dozens of other software companies to distribute the previously undocumented Rapuncel infostealer. The campaign also delivers a Microsoft-signed kernel driver capable of disabling a large number of antivirus and EDR products.

Source: BleepingComputer