All news in category “Security Advisory and Patch Watch”

🛡️ Varonis disclosed a vulnerability called RovoBlast in Atlassian's enterprise AI assistant, Rovo, which allowed a crafted URL parameter to seed attacker instructions into an authenticated session. The assistant's ResearchAgent could then browse the web and post retrieved internal data externally, enabling data leakage with a single click. Atlassian has since patched the issue; Varonis urges restricting connectors, disabling browsing agents, and monitoring agent activity.

🔐 Three research teams disclosed attacks that bypass passkey protections without breaking FIDO cryptography. SpecterOps showed Windows-exposed signatures chained through Microsoft Entra ID to impersonate privileged users. Unit 42 demonstrated methods to recover synced passkey private keys in Chrome's Google Password Manager, and Dirk-jan Mollema showed malware in a signed-in Windows session could use a Windows Hello for Business key without a fresh PIN. Vendors issued patches and mitigations with differing impacts.

🔒 The U.S. Cybersecurity and Infrastructure Security Agency (CISA) warns that threat actors are actively exploiting a critical command injection vulnerability in Progress Kemp LoadMaster. The flaw, tracked as CVE-2026-8037, allows unauthenticated attackers to execute arbitrary commands on unpatched appliances via unsanitized API inputs. Progress released patches in June for affected GA and LTSF versions, and CISA has directed federal agencies to remediate within three days.

🔒 Researchers have identified malicious Visual Studio Code extensions named Solidity Pro that evolved from fetching encrypted payloads to a full-featured information stealer. The extensions, distributed under names like helper-beeps.solidity-pro and web3devtoolsx.solidity-pro, harvest browser profiles, crypto wallets, SSH keys, API tokens, and more, then exfiltrate data via a Telegram bot. The malware uses heavy obfuscation, staged clean releases, and randomized delayed activation to evade detection and marketplace review.

🛡️ Two security teams found ways to make Atlassian's Rovo assistant exfiltrate data a signed-in user can access. One method used a malicious file with hidden instructions to induce Rovo to gather Jira or Confluence content and send it to an attacker-controlled URL; PromptArmor disclosed this on August 5, 2026 and its remediation status after publication is unconfirmed. The second, dubbed RovoBlast by Varonis, preloads attacker instructions via a rovoChatPrompt URL parameter so a single click from an authenticated user could cause data to be sent out; Atlassian fixed this server-side on July 8, 2026. Both issues rely on data the signed-in user can reach, and administrators can limit exposure by restricting which apps and groups can use Rovo and tightening connector permissions.

🔒 New research shows HTML and CSS can escape email message boundaries to interfere with webmail UIs across major providers. PortSwigger researcher Gareth Heyes presented proof-of-concept chains at Black Hat USA 2026 targeting Outlook, Gmail, Fastmail, Proton Mail, Yahoo Mail, and AOL Mail. The techniques can capture passwords, leak tokens, hijack UI actions, and manipulate AI-connected tools; some PoCs remained public as of August 8.

🔒 N‑able has issued Hotfix 2 for N‑central after detecting active exploitation of a recently disclosed RMM server vulnerability (CVE-2026-18577) first observed on July 31, 2026. The company says Hotfix 2 supersedes Hotfix 1 and provides additional hardening; on-prem customers must update to 2026.3.1.10 immediately. A limited set of customers were affected, and N‑able published IoCs plus a custom service template to scan Windows endpoints, while cautioning that results are not a guarantee of full remediation.

🔒 CISA has added a critical command injection vulnerability in Progress Kemp LoadMaster (CVE-2026-8037, CVSS 9.6) to its Known Exploited Vulnerabilities catalog after reports of active exploitation. The flaw, rooted in improper input handling in an escape_quotes() function, allows unauthenticated attackers to execute arbitrary commands on affected appliances. Agencies are urged to apply patches immediately under BOD 26-04 to mitigate ongoing attacks.

🔒 WordPress patched a pre-auth reflected XSS in the login screen (CVE-2026-64638) that requires no attacker privileges and can execute JavaScript when a crafted username reaches the failed-login page. Researchers at pwn.ai demonstrated chaining the XSS to PHP code execution if an Administrator interacts with an attacker-controlled page, and WordPress issued fixes on August 6 across supported branches.

🛡️ A long-standing use-after-free bug in Linux's SCTP implementation, tracked as CVE-2026-64564 and called SCTPhantom, can be exploited to achieve local privilege escalation and, according to Tencent Zhuque Lab, to escape containers and reach the host. Stable kernel fixes (7.1.6, 6.18.42, 6.12.101 and 6.6.148) were released on August 3. Systems with reachable SCTP should apply vendor updates or disable the module if unused.

🔒 Researcher Malcolm Stagg disclosed NatJack, a class of attacks that manipulates NAT connection state to hijack TCP sessions, spoof DNS responses, expose mapped ports, and exhaust NAT tables. Presented at Black Hat USA 2026, the work found vulnerable behaviors across multiple implementations including Windows and Linux, with two CVEs assigned: CVE-2026-56181 and CVE-2026-63913. Mitigations include applying vendor updates, segregating untrusted workloads, encrypting internal traffic, and employing IP Source Guard where possible.

🔒 Entra ID researcher Dirk-jan Mollema demonstrated that malware running in a signed-in Windows session can silently invoke the victim's Windows Hello for Business key to authenticate to Microsoft Entra ID. The technique lets an attacker obtain tokens, register devices, and gain long-term cloud access without extracting private keys, recovering PINs, or prompting biometrics on TPM-backed systems. Mollema published PoC scripts and recommends hunting for Hello sign-ins with empty device IDs while noting potential false positives.

🛡️ At Black Hat USA 2026, researcher Malcolm Stagg disclosed NatJack, a class of attacks that manipulates NAT connection tracking tables to hijack TCP connections, poison DNS, and cause DoS without needing IP spoofing or Layer 2 access. Testing across 32 products revealed vulnerabilities in every implementation examined. Vendor responses varied from patches and CVEs to arguments that the issues reflect design limitations rather than security flaws. Stagg recommended monitoring NAT tables, enabling source IP protections, segmenting untrusted traffic, and disabling loose connection tracking modes as mitigations.

🛡️ Researchers at MIT CSAIL disclosed a new CPU side-channel exploit called TONTOU that bypasses neutralization-based Spectre v2 mitigations on Intel and AMD processors. They developed an Interrupt Injection technique to re-poison branch predictors after mitigation cleaning and demonstrated leaking kernel memory, including /etc/shadow hashes, from Linux machines. The team presented results at Black Hat USA and will publish further details at USENIX Security 2026.

🔒 Zapscape (CVE-2026-64561) is a Linux KVM/x86 shadow-MMU flaw that can let an attacker with kernel privileges in an L1 guest escape KVM isolation and run code on the host. Disclosed by researcher Hyunwoo Kim, the issue is a stale-root ordering bug causing a use-after-free during page-fault handling when nested virtualization is exposed. The upstream fix has been merged; administrators should update kernels or vendor packages that backport the patch.

🔒 Cisco issued patches for multiple critical vulnerabilities in Catalyst SD‑WAN and IOS XE Software discovered during an internal security review. The flaws—ranging from improper input validation and access control to command injection—affect many releases and have been fixed across several patched versions. Cisco noted these were found during testing, including use of frontier AI models, and are not known to be actively exploited, urging customers to update promptly.

🔒 Researchers from MIT CSAIL discovered INTERRUPT INJECTION, a technique that times interrupts to re-poison the branch predictor between a processor's sanitization and kernel use, bypassing Spectre v2 mitigations. On AMD Zen 2 with Linux 6.14 and default protections, their exploit read kernel memory at ~5.47 B/s and retrieved /etc/shadow in multiple trials. AMD issued bulletin AMD-SB-7061 and plans patches; Intel currently deems mitigation unnecessary. The disclosure highlights gaps in detection and reporting for deployed fixes.

🔒 Researchers disclosed that features in Apple's WebKit can bypass configured proxies and reveal users' real IP addresses, affecting iCloud Private Relay. The flaw stems from DNS prefetching, WebAuthn related origin requests, and WebTransport, which send traffic outside the relay. A PoC site demonstrates the leak, and Apple says it is investigating while the issue also impacts macOS and other WebKit-based browsers.

🛡️ Security advisories from Amazon Web Services, Google, and Vercel detail vulnerabilities in agent harnesses that allowed caller-supplied or forged tool-call data to reach execution without a verifying model turn. AWS patched its managed Bedrock AgentCore service, Google fixed ADK for Python in v2.5.0, and Vercel released fixes for its Codex and OpenCode harness packages. Each vendor's path differed in attack surface and required conditions, and mitigations focus on validating and binding tool invocations to authenticated model events.

🔒 JetBrains TeamCity on-premises installs are affected by CVE-2026-63077, a deserialization flaw enabling unauthenticated remote code execution via the agent polling protocol. An attacker can bypass authentication and run OS-level commands with the TeamCity process privileges, risking exposure of data, credentials, and build integrity. CISA reports active exploitation and urges immediate patching; federal agencies must remediate by August 8, 2026 under BOD 26-04.