The day's cybersecurity headlines from public sources, in one place. Search them, filter by theme or CVE, or show only vulnerabilities CISA has confirmed are being exploited. Every headline links to the original.
For my own read on what the news means for your business, see Insights.
-
Fake LastPass Authenticator GitHub repositories push the new Rapuncel infostealer
An ongoing campaign uses SEO-optimized GitHub repositories impersonating well-known software vendors to distribute a previously undocumented infostealer called Rapuncel. The repositories rank in search results, which is how victims arrive.
-
WordPress Click2Shell flaw forces theme installs from a crafted link
A crafted link opened by a logged-in WordPress administrator can install a theme from the official WordPress.org directory without anyone clicking Install. Security firm pwn.ai named the chain Click2Shell. On its own the flaw only installs a legitimate theme of the attacker's choosing and leaves…
-
Critical Orkes Conductor flaw under active exploitation, roughly 1,300 attempts blocked in two days
CVE-2026-58138 is an unauthenticated remote code execution flaw (CVSS 9.8) in Orkes Conductor, the open source workflow and microservice orchestration framework. An attacker submits a malicious workflow definition with JavaScript or Python expressions to the workflow API, and because the GraalVM…
-
FBI and Coast Guard boarded two hacked oil tankers headed toward the US coast
Federal teams boarded two foreign-flagged tankers in the Gulf of Mexico on August 21 and August 24 after indications that attackers had reached shipboard networks. One was the VL Prosperity, a Liberian-flagged supertanker carrying roughly 2.3 million barrels. Rear Adm. Amy Grable of Coast Guard…
-
Microsoft patches a CVSS 10.0 Azure AI Foundry flaw and 17 other cloud and AI vulnerabilities
CVE-2026-85889 is a missing-authentication flaw in Azure AI Foundry (also called Microsoft Foundry) that allows an unauthorized attacker to elevate privileges over the network. Microsoft rates it 10.0 and says no customer action is required because the fix was applied service-side. There is no…
-
Plugin4Shell lets repository owners swap pinned plugin code in four AI coding agents
Air Security disclosed that four widely used AI coding agents fetch plugin code pinned to a commit hash but never verify that what they received matches that hash. On a code host that permits a branch named to look like a commit hash, the plugin repository owner can substitute malicious code for…
-
CISA adds three Linux kernel flaws to the KEV catalog in a single day
CISA issued two separate KEV alerts on the same day. The first added CVE-2025-39964, a race condition in AFALG socket handling (CVSS 7.8) that lets a local attacker crash the system or corrupt cryptographic results, and CVE-2026-53266, an out-of-bounds write in the ebtables SNAT ARP rewrite path…
-
Public exploit code released for four Linux kernel local-root flaws
Researcher Asim Manizada published working exploit code for four Linux kernel vulnerabilities he calls DirtyAH6, TUNderflow, PPPoEject, and DiagSpill. Each gives a local user root. Kernel maintainers fixed all four over the past several weeks under a coordinated hold with distributions, so a…
-
OpenAI says its models searched GitHub for leaked API keys during training
OpenAI disclosed that during training, its models searched GitHub for leaked API keys. The disclosure is part of the same transparency push behind the misalignment framework above.
-
Research shows AI agents can retrain their own models mid-task, leaking secrets and erasing refusals
Researchers demonstrated that AI agents with sufficient tool access can initiate fine-tuning or retraining of their own underlying models during a task, with two consequences: secrets present in the working context can be baked into model weights and later leaked, and safety refusals can be…
-
OpenAI discloses six new model misalignment incidents and a reporting framework
OpenAI disclosed six instances of unexpected or concerning model behavior from the past six months, including unauthorized file uploads, models following self-generated instructions, hiding mistakes, and leveraging exposed API keys. These are separate from the previously disclosed misaligned…
-
CISA releases guidance on deploying cyber decoys for critical infrastructure
CISA published guidance on deploying cyber decoys, meaning honeypots, canary tokens, and deception assets, as a detection layer for critical infrastructure operators. The framing is detection capability for organizations that cannot achieve full visibility through monitoring alone.
-
US seizes NightmareStresser DDoS-for-hire domains
DOJ announced court-authorized seizure of nightmare-stresser.com and nightmarestresser.org, one of the longest-running booter services, linked to hundreds of thousands of DDoS attacks. The seizure was executed by the FBI Anchorage Field Office with the US Attorney for the District of Alaska and…
-
CISA retires the weekly vulnerability bulletin and shifts federal agencies to a three-day exploit-evidence clock
CISA is discontinuing its weekly vulnerability bulletin on September 28 in favor of risk-based prioritization under Binding Operational Directive 26-04. The directive replaces severity-score patching for federal civilian agencies with a decision tree built on four signals: whether the asset is…
-
ISC patches 14 flaws in BIND 9, including an unauthenticated crash over DNS-over-HTTPS
ISC released BIND 9.20.29 and 9.21.26 to fix fourteen flaws disclosed September 16. One affects any BIND server answering DNS-over-HTTPS, where an unauthenticated sender can crash the named process with a single request carrying an invalid SIG(0) signature if the sender closes the connection…
-
Gyazo breach exposes 23.62 million user records and 490 million image metadata records
Helpfeel disclosed Wednesday that a breach of its Gyazo image-sharing service exposed roughly 23.62 million user records including email addresses and password hashes, plus about 490 million image metadata records, mostly for images from January 2019 or earlier. The exposed metadata includes the…
-
Revolut breach: hackers posed as an Italian government agency for five months, now demanding $3 million
An attacker compromised an Italian Ministry of the Interior employee account through an infostealer infection, then used that government email address to send fraudulent legal requests to Revolut Bank UAB in Lithuania for roughly five months. Revolut, obligated to respond to law enforcement…
-
Ransomware attacks on manufacturers up 40 percent, with half the attacks from groups that did not exist two years ago
Black Kite counted 1,183 new manufacturing ransomware incidents in the first seven months of 2026, a 40 percent rise over the same period in 2025. Half the attacks came from groups that did not exist two years ago, and one new group, The Gentlemen, accounted for 12 percent of this year's attacks…
-
Iran-linked Handala Hack tied to HEAVYGRAM Telegram backdoor
Group-IB attributes a Telegram-based surveillance backdoor called HEAVYGRAM and a Delphi loader called CRUDEEXCLUDE to the Iran-linked hacktivist persona Handala Hack. HEAVYGRAM supports remote command execution, discovery, exfiltration including Telegram session files, screenshot capture, DLL…
-
China-aligned FamousSparrow deploys SparroWocky backdoor across Latin America
ESET reports FamousSparrow, which overlaps with Earth Estries and Salt Typhoon, has replaced SparrowDoor with a new modular C++ backdoor called SparroWocky in attacks on government organizations across multiple Latin American countries since at least August 2025. Researchers note strong…
-
New RatHat Android malware uses an AI subsystem to automate device control
RatHat is a newly discovered Android remote access trojan with an AI-powered subsystem that helps operators navigate a compromised device remotely rather than requiring manual step-by-step control. The design point is scale, since AI-assisted navigation lets one operator work many infected devices.
-
Critical Docker Sandboxes flaw lets malicious guest code read and modify macOS host files
CVE-2026-77179 lets code running inside a Docker Sandboxes virtual machine on macOS escape the shared project directory and read or change files anywhere the host account can reach. It affects versions 0.28.0 up to 0.42.0 on macOS and was fixed in 0.42.0 on September 7. Docker Sandboxes exists to…
-
Critical Unbound DNSSEC validator flaw allows remote code execution via a malicious DNS zone
Every Unbound DNS resolver release before 1.26.1 carries a critical heap overflow in the DNSSEC validator, tracked as CVE-2026-81642. An attacker who controls a malicious zone and gets a vulnerable resolver to query it can trigger remote code execution. Unbound 1.26.1 fixes it along with eight…
-
Brevo supply-chain attack injected ClickFix scripts on up to 100,000 customer sites
Attackers used a long-lived Cloudflare API key with full account permissions, hardcoded in Brevo application source code, to create a malicious Cloudflare Worker that rewrote responses at the CDN edge for about five and a half hours on September 14. Because the rewrite happened at the edge and…
-
Critical Check Point management server flaw lets unauthenticated attackers run code as root
CVE-2026-91843, rated 9.8 by Check Point, is a stack overflow in the pre-authentication login path of Check Point Security Management and Log Servers. Censys says the overflow is triggered by a login request carrying a very long username, and the vulnerable path runs only through the Trusted…