Drupal has shipped the highly critical core security release teased by PSA-2026-05-18. The flaw lets attackers achieve remote code execution on Drupal sites running PostgreSQL backends. Fixed versions are 11.3.10, 11.2.12, 11.1.10, 10.6.9, 10.5.10, and 10.4.10. The releases for supported branches also pull in upstream Symfony and Twig security fixes, making the upgrade essential even on MySQL deployments. Best-effort manual patches are available for end-of-life Drupal 9.5 and 8.9. Drupal 7 is not affected. The Drupal Security Team had warned that working exploits could follow within hours of disclosure, so administrators should patch now.
HiddenLayer has disclosed a maximum-severity unauthenticated remote-code-execution vulnerability, CVE-2026-45829, in ChromaDB's Python FastAPI server. ChromaDB is one of the most popular vector databases backing retrieval-augmented-generation pipelines, with about 14 million monthly PyPI downloads. A vulnerable endpoint marked as authenticated lets an attacker embed model settings before authentication is checked, so a crafted request makes ChromaDB load a malicious model from Hugging Face and execute it locally. The auth check fires only after the payload has already run. The bug was introduced in 1.0.0 and was still present in 1.5.8. HiddenLayer's Shodan sweep shows ~73% of internet-exposed Chroma instances are vulnerable.
Drupal is releasing an emergency core security update on May 20 between 17:00 and 21:00 UTC. Pre-disclosure advisory PSA-2026-05-18 rates the issue 'highly critical' (20 of 25 on Drupal's scoring) and notes access complexity 'None' and authentication 'None' - meaning the underlying flaw is unauthenticated and easy to trigger. Patches will land for the supported 11.3.x, 11.2.x, 10.6.x, and 10.5.x branches, plus emergency patches for EOL 11.1.x and 10.4.x. Drupal 7 is not affected. Drupal 8 and 9 will only get best-effort manual patch files. The Drupal Security Team warns working exploits may follow within hours of disclosure.
Researchers have disclosed a chain of vulnerabilities in SEPPmail Secure Email Gateway that lets an attacker turn unauthenticated web requests into remote code execution by inflating the SEPPMaillog file past its 10,000 KB limit, which forces newsyslog to rotate logs and signal syslogd to reload its configuration. Combined with the other flaws in the chain, the attacker reads all mail traffic on the appliance and persists indefinitely. SEPPmail has patched CVE-2026-44128 in version 15.0.2.1, CVE-2026-44126 in 15.0.3, and the rest in 15.0.4. The disclosure follows last month's CVE-2026-27441 (CVSS 9.5) OS command-execution fix in the same appliance.
Recorded Future News has connected last summer's three-hour POST Luxembourg outage - which took down landline, 4G, and 5G networks across the country and left residents unable to dial emergency services - to a zero-day in Huawei enterprise routers running VRP. Specially crafted network traffic merely passing through caused the routers to enter a continuous restart loop. Luxembourg's prosecutor concluded no one had targeted Luxembourg specifically; the data was just transit traffic. Huawei has not assigned a CVE for the bug and routes its enterprise advisories through a restricted customer portal rather than publicly, leaving operators with little ability to track exposure.
A working proof-of-concept exploit for a recently patched Linux kernel local privilege escalation is now public. Researchers at V12 found the bug in May and were told it had already been fixed in the mainline kernel on April 25, matching CVE-2026-31635 per Tharros analyst Will Dormann. The flaw is a missing copy-on-write check in rxgk_decrypt_skb, the kernel routine that decrypts RxGK packets for the Andrew File System. Exploitation requires CONFIG_RXGK, limiting impact to leading-edge distros like Fedora, Arch Linux, and openSUSE Tumbleweed. DirtyDecrypt joins Dirty Frag, Fragnesia, and Copy Fail in a recent wave of Linux LPE disclosures.
Researchers at Cyera have disclosed a chain of four vulnerabilities in OpenClaw, an open-source autonomous AI agent platform that Nvidia and Tencent have built enterprise products on top of. The chain - CVE-2026-44112 (CVSS 9.6), CVE-2026-44113, CVE-2026-44115, and CVE-2026-44118 - lets an attacker who can influence the agent's input (through a malicious plugin, prompt injection, or compromised tool output) break out of the OpenShell sandbox, read environment-stored API keys, elevate to owner-level privileges, and write persistent backdoors. Each step looks like normal agent behavior. Shodan and Zoomeye between them counted 65,000 to 180,000 public OpenClaw instances earlier in May. All flaws are fixed in OpenClaw 2026.4.22.
A wave of critical patches landed across enterprise vendors. Fortinet shipped fixes for two unauthenticated code-execution flaws (CVE-2026-44277 in FortiAuthenticator, CVE-2026-26083 in FortiSandbox / FortiSandbox Cloud / FortiSandbox PaaS, both CVSS 9.1). SAP patched a 9.6-rated SQL injection in S/4HANA and a missing-auth check in SAP Commerce that allows unauthenticated code execution. Ivanti Xtraction got a fix for arbitrary file read and write. Broadcom patched a VMware Fusion macOS local-privilege-escalation (CVE-2026-41702). And the n8n automation platform shipped five CVSS 9.4 issues, including XML-driven prototype pollution that authenticated workflow editors could turn into RCE.
A researcher who goes by Chaotic Eclipse has dropped working proof-of-concept code on GitHub for a Windows local privilege escalation that gives SYSTEM access on fully patched Windows 11 Pro and Windows Server 2025. The bug lives in the Cloud Filter driver cldflt.sys and is, the researcher says, the same flaw Google Project Zero reported to Microsoft as CVE-2020-17103 in 2020, which Microsoft said it fixed in December 2020. The original Google PoC works unmodified. May 2026 Patch Tuesday updates do not stop it. The same researcher has dropped several other Windows zero-days in recent weeks, all of which were quickly seen in real attacks.
The 18-year-old heap overflow in NGINX's rewrite module, CVE-2026-42945, disclosed last week as part of the 'Rift' bug cluster, is now seeing real exploitation attempts. AI-native security firm VulnCheck says its honeypot networks have caught attackers probing the flaw, though the goal of the campaigns is not yet clear. The vulnerability lets an unauthenticated attacker crash NGINX worker processes by sending crafted HTTP requests. Turning that crash into remote code execution requires the target host to have Address Space Layout Randomization (ASLR) disabled, which is uncommon by default, but the worker-crash denial-of-service is exploitable on its own and rated urgent.
TrustStrike Labs