All news with #vulnerability disclosure tag

🚨 Over 1,300 Internet-exposed Microsoft SharePoint servers remain unpatched against CVE-2026-32201, a spoofing vulnerability Microsoft fixed in its April 2026 Patch Tuesday. The flaw affects SharePoint Enterprise Server 2016, SharePoint Server 2019, and SharePoint Server Subscription Edition and was flagged as a zero-day exploited in the wild. Fewer than 200 systems have been patched since the update; organizations should apply Microsoft's fixes or recommended mitigations immediately.

⚠️ Anthropic's reported Claude Mythos marks a shift: AI is compressing attack timelines by accelerating vulnerability discovery, exploit development, and multi-step attack planning. Attackers can now run malware, phishing, and vulnerability exploitation in parallel, reducing time to compromise and widening exposure. This trend demands prevention-first controls and real-time detection to identify and remediate gaps earlier, limiting impact.

⚠️ Silex Technology released updates addressing multiple serious vulnerabilities in SD-330AC and AMC Manager that could permit remote code execution, denial-of-service, or unauthenticated configuration changes. Affected versions include SD-330AC ≤ 1.42 and AMC Manager ≤ 5.0.2; vendor fixes are SD-330AC firmware 1.50+ and AMC Manager 5.1.0+. CISA notes CVSS scores up to 9.8 and recommends applying vendor updates and interim mitigations such as disabling HTTP/HTTPS for impacted functions, setting web-interface passwords, and disabling SNMP.

⚠️ CISA added eight vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog, citing evidence of active exploitation and highlighting three flaws in Cisco Catalyst SD-WAN Manager. The list includes high-impact issues such as CVE-2025-32975 (Quest KACE SMA, CVSS 10.0) and authentication, path traversal, and XSS flaws in PaperCut, TeamCity, Kentico, and Zimbra. CISA noted prior ties of CVE-2023-27351 to Lace Tempest and recent Arctic Wolf telemetry on KACE abuse; Cisco confirmed active exploitation of two SD-WAN flaws in March 2026. Federal civilian agencies are urged to remediate the three Cisco vulnerabilities by April 23, 2026, and the remaining flaws by May 4, 2026.

⚠️ A critical vulnerability (CVE-2026-5760) in SGLang allows remote code execution via specially crafted GGUF model files. The flaw targets the /v1/rerank endpoint, where a malicious tokenizer.chat_template containing a Jinja2 SSTI payload is rendered using an unsandboxed jinja2.Environment(), enabling arbitrary Python execution. Researcher Stuart Beck reported the issue to CERT/CC, which recommends replacing jinja2.Environment() with ImmutableSandboxedEnvironment to mitigate the risk. No patch was obtained during coordination.

⚠️ A critical remote code execution vulnerability has been disclosed in protobuf.js, the widely used JavaScript implementation of Google's Protocol Buffers, caused by unsafe dynamic code generation that concatenates schema-derived identifiers into functions. An attacker who can supply or influence schemas can inject arbitrary JavaScript into a generated Function() call, which executes when the crafted schema is processed. Maintainers and Endor Labs urge immediate upgrades to patched releases and recommend treating schema-loading as untrusted while auditing transitive dependencies.

⚠️ Maintainers of Thymeleaf released a patch addressing a critical Server-Side Template Injection (SSTI) vulnerability, tracked as CVE-2026-40478, that allows unauthenticated attackers to execute expressions and run code. The flaw bypasses Thymeleaf’s sandbox protections by exploiting control characters in expressions and improper class restrictions. All versions prior to 3.1.4.RELEASE are affected, there is no workaround, and organizations should upgrade immediately.

🔎 Unit 42 observed automated scans targeting CVE-2023-33538 in several end-of-life TP‑Link routers (TL‑WR940N, TL‑WR740N, TL‑WR841N). Payloads resembled Mirai-like botnet binaries and attempted to download and execute an arm7 ELF, but in-the-wild attempts were flawed and generally failed. Emulation and reverse engineering confirmed a real command-injection flaw in the ssid1 parameter that reaches a system shell, but successful exploitation requires web authentication (default credentials like admin:admin remain a practical risk). TP‑Link lists the devices as EOL with no patches; Unit 42 recommends replacing affected units and avoiding default credentials while using layered protections.

🔒 Cisco Talos disclosed a use-after-free flaw in Foxit Reader (TALOS-2026-2365 / CVE-2026-3779) exploitable via malicious PDF JavaScript, and six vulnerabilities in LibRaw including heap-based buffer overflows and integer overflows across multiple CVEs. All issues were patched by vendors following Cisco’s disclosure policy. Administrators should apply vendor updates and deploy Snort rules from Talos to detect exploitation.

🛡️ This week's bulletin highlights both legacy and emerging threats, including a published Microsoft Defender privilege escalation exploit (RedSun) and a 17‑year‑old Excel RCE (CVE‑2009‑0238) newly added to CISA's KEV. Incidents range from a Zerion hot-wallet compromise (~$100K stolen through AI‑enabled social engineering) to a fake macOS Ledger app that drained about $9.5M. Researchers also disclosed novel C2 frameworks, a WordPress plugin supply-chain backdoor affecting 180k+ installs, and a surge in SonicWall/FortiGate brute-force probing. The collection underscores the need to patch promptly, validate app-store integrity, rotate credentials, and audit third-party dependencies.

🔍VulnCheck's analysis indicates Anthropic's controlled-access Project Glasswing has only one publicly attributable CVE: CVE-2026-4747, a FreeBSD NFS remote code execution flaw described as autonomously identified and exploited. Researcher Patrick Garrity reviewed the CVE database and found 75 records mentioning Anthropic, but only 40 credited to its researchers and a single CVE tied explicitly to Glasswing. Industry observers warn that public attribution may understate the model's potential, and Anthropic plans a fuller accounting by July 2026.

🔒 At VulnCon26 in April, Lindsey Cerkovnik of CISA urged that AI firms like OpenAI and Anthropic be more directly represented in the CVE program to help manage a surge in reported vulnerabilities. She warned that new AI tools both accelerate discovery of valid flaws and generate lower-value noise, putting pressure on disclosure workflows. Recent vendor developments — Anthropic’s Mythos Preview and OpenAI’s GPT-5.4-Cyber — illustrate how automated research is already changing the threat landscape. Cerkovnik said CVE funding is secure and the program remains a CISA priority.

⚠️ Two high-severity vulnerabilities in Composer's Perforce VCS driver (CVE-2026-40176, CVSS 7.8; CVE-2026-40261, CVSS 8.8) can enable arbitrary command injection when processing a malicious repository configuration or a crafted source reference. The issues affect releases prior to 2.9.6 and 2.2.27 and are fixed in those versions; users should upgrade immediately. If you cannot patch, inspect composer.json files for Perforce fields, restrict repositories to trusted sources, and avoid dist-preferred installs. Composer reported no evidence of public exploitation and disabled Perforce metadata publishing on Packagist.org as a precaution.

⚠️ A critical remote code execution vulnerability, tracked as CVE-2025-0520 (aka CNVD-2020-26585), is being actively exploited against unpatched instances of ShowDoc. The flaw is an unrestricted, unauthenticated file upload caused by improper file-extension validation, allowing attackers to deploy PHP web shells and execute arbitrary code. The bug was fixed in ShowDoc 2.8.7 (October 2020) and the project now ships as version 3.8.1, but researchers observed an exploit dropping a web shell on a U.S.-based honeypot and note more than 2,000 internet-facing instances, most located in China. Administrators should upgrade immediately and scan for signs of compromise.

🔍 Anthropic's new Claude Mythos Preview and its Project Glasswing effort have focused industry attention on AI-driven cyberattack capabilities. Anthropic says it will not release the model publicly, citing the risk that it can automatically generate operational exploits, and is running the model against public and proprietary code to find and patch vulnerabilities before they can be weaponized. The announcement produced substantial PR impact, prompting rival vendors to echo similar caution. Security observers note defenders still hold an advantage—finding flaws is easier than turning them into attacks—but that margin is shrinking as models improve.

🔔 Emergency updates address a critical PDF zero‑day in Adobe Acrobat Reader (CVE-2026-34621, CVSS 8.6) that executes malicious JavaScript when specially crafted documents are opened. The report also highlights Anthropic's Mythos being used as an exploit-generation engine, state-linked interference with infrastructure, and research showing telecom optical fibers can be abused for acoustic eavesdropping. Prioritize patching, credential hygiene, and detection for fileless and AI-driven attacks.

🔔 CISA added seven vulnerabilities to the Known Exploited Vulnerabilities (KEV) Catalog after evidence of active exploitation, affecting Microsoft, Adobe, and Fortinet products. The CVEs cover insecure library loading, use‑after‑free, deserialization, out‑of‑bounds read, link following, SQL injection, and prototype pollution. Under BOD 22‑01, Federal Civilian Executive Branch agencies must remediate KEV entries by required dates, and CISA urges all organizations to prioritize timely remediation as part of routine vulnerability management.

⚠️ A critical pre-auth remote code execution (RCE) in Marimo (CVE-2026-39987) permits unauthenticated access to an interactive shell via the /terminal/ws WebSocket endpoint in versions 0.20.4 and earlier. Sysdig observed exploitation beginning within 10 hours of the public disclosure, with attackers quickly harvesting .env files, cloud credentials and SSH keys. Marimo released v0.23.0 to patch the issue; users should upgrade immediately, restrict external access, monitor WebSocket connections, and rotate any exposed secrets.

⚠️ Researchers disclosed a critical remote code execution flaw in Apache ActiveMQ Classic that remained undetected for 13 years and can allow arbitrary system command execution. Tracked as CVE-2026-34197 with a CVSS score of 8.8, the bug affects Classic releases before 5.19.4 and 6.0.0 through 6.2.3; fixes were released in 5.19.4 and 6.2.3. Administrators should apply the updates, review Jolokia access controls, and inspect broker logs for indicators of compromise.

⚠ A critical arbitrary file upload vulnerability has been identified in the Ninja Forms – File Upload Plugin for WordPress, impacting versions up to 3.3.26 and rated CVSS 9.8. The flaw allows unauthenticated attackers to upload malicious files (including .php), bypass validation, and achieve remote code execution. Wordfence validated the report after it was disclosed on January 8, 2026, and the developer issued a complete patch in version 3.3.27 on March 19; administrators should update immediately.