The PHP Foundation has released security updates to address multiple critical security flaws that could be exploited to achieve remote code execution on affected deployments, including CVE-2026-115560. The release of the patch follows an advance notice in which the project maintainers alerted users to multiple critical vulnerabilities affecting all supported PHP versions under specific
Cybersecurity News and Vulnerability Aggregator
Cybersecurity news aggregator
treemd <(curl -sL https://allsec.sh/md) (as Markdown) Top Cybersecurity Stories Today
Latest
Yahoo Finance: Palo Alto Networks CEO just dropped a truth bomb on the network outages sweeping through major companies
Link: https://finance.yahoo.com/technology/article/palo-alto-networks-ceo-just-dropped-a-truth-bomb-on-the-network-outages-sweeping-through-major-companies-101129581.html The title is as in the original article - it is not editorialized. The tl;dr is: > "I want to be clear, there are two distinct parallel effects that are driving this environment. The first effect is a massive surge in vulnerabilities and patches being deployed. We're seeing critical and severe vulnerabilities discovered nearly seven times the monthly average of last year. These discoveries are happening at a speed and intensity which is hard for IT teams to cope with," > > [...] > > "The second effect is the rise of AI-assisted cyberattacks. Bad actors are using open-weight models and new agentic frameworks to find and exploit vulnerabilities at unprecedented speed and scale. New agentic frameworks like ARTEX are enabling bad actors to turn simple instructions into autonomous, multi-stage attacks, allowing a handful of operators to compromise dozens of targets simultaneously." The statements by themselves probably are nothing new for folks out here, but the point is that the message and the unintended consequences of _mOaR aI eVeRyWhErE_ are starting to finally bubble up to the very top.
The company’s alternate-universe internet serves as a massive sandbox that allows the Chrome team to properly train AI browsing agents—without the risk of them going rogue.
Am I the only one thinking this rhetoric is irresponsible. Also, who in their sane mind wears an EC Council shirt? [https://www.foxnews.com/video/6408568867112](https://www.foxnews.com/video/6408568867112)
The PHP Foundation has released security updates to address multiple critical security flaws that could be exploited to achieve remote code execution on affected deployments, including CVE-2026-115560. The release of the patch follows an advance notice in which the project maintainers alerted users to multiple critical vulnerabilities affecting all supported PHP versions under specific
Visitors to government election websites surged 363% on November 3, the date of the 2026 midterm elections in the United States. Across Cloudflare's election protection programs, we also detected a significant shift in the types and timing of cyberattacks throughout 2026. Unlike what we saw in other recent elections, attackers focused on gaining access to internal systems and data, rather than knocking websites offline. Government election websites faced website vulnerability attempts 180 times more often than denial-of-service (DDoS) attacks, and at roughly ten times the rate of Cloudflare's broader customer base. We also found evidence of at least one attempt to exploit a recently disclosed critical PHP vulnerability on a state government voting website in the Western United States. These attacks peaked in late October, several weeks before the election. Vulnerability exploit attempts climbed steadily throughout 2026: from 200,000 per day in January, to 2 million per day in October. However, on Election Day in the U.S., Cloudflare detected no escalation in attack volume and zero major DDoS incidents across protected election sites, marking a sharp contrast to other recent elections, such as in Moldova last year. Our data regarding cyberattacks on election institutions came from organizations participating in Cloudflare's free cybersecurity programs: the Athenian Project , Cloudflare for Campaigns , and Project Galileo . Together, these initiatives helped protect more than 1,400 election-related websi
What External Network Penetration Testing Covers External network penetration testing evaluates an organization’s internet-facing infrastructure from the position of an attacker who holds no credentials and no internal access, working from publicly available information the same way a real adversary would. That starting position is what separates it from every other testing discipline, since the […] The post External Network Penetration Testing: Enterprise Methodology and What’s Included appeared first on Synack .
One of the early targets of PHP2Shell exploitation was the firewall. On October 28, Palo Alto Networks confirmed that attackers were exploiting the two PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619, against PAN-OS management interfaces exposed to the internet. Both flaws let anyone execute code on the web server without authentication. The disclosure itself went a little wrong.
OPTIMJACK: the iPhone exploit router behind the BBC and Zoom vendor-script poisoning is rented out by channel code
The router that Kissmetrics, StorifyMe, minute-ly and Optimizely scripts loaded on customer pages in November is a service with customers. Each page that loads it passes a channel code, and the code tells the operator whose visitor the phone is. **The house channel.** The vendor scripts carried `02111110`. Censys found the same code as the kit's default and read it as naming no operator. It is also `DEFAULT_CH` in the operator's own lure template (a fake TRON "Energy Rental" shop) and the channel set in the clipper config on Optimizely's files, so we read it as the operator's own channel. **The renters.** We hold 31 other channel codes, in use from 31 August 2026. Nine channels had their router host and lure registered together, mostly minutes apart; for `0649c29a` both were registered at 17:30 UTC on 10 November at two different registrars. Renters include: - BranchHop, a fake-verification (ClickFix) kit that added the router to its stage on 33 hacked sites. iPhones got the router; Windows visitors kept getting a PowerShell command. - A LenAI operator, on eight hacked sites. - A gambling network, fake Trust Wallet sites and a dating site. **The console.** Router servers also serve an "AIGO agent backend" login dressed as cloud backup. Hierarchy is agent, user, device; agents apply and "Headquarters" approves them. It lists each phone's wallets, keystores, Keychain, messages and photos and sends Telegram alerts. **Routing.** `boot.js` returns an empty file to desktops. iOS 13 to 17 goes to Coruna, 18.4 to 18.7.2 to DarkSword, and 18.7.3 and later to GhostWave, a DarkSword fork the kit's notes grade as not production-ready. **Delivery.** The vendor scripts loaded the router from an address of linkforge, a seller of hidden backlinks on hacked sites. On 9 November linkforge's script swapped its ad code for the router. **Status.** Registrars put the vendor loader domains on hold on 17 November and `linkforge[.]cc` on 18 November. At 22:10 UTC on 18 November nine of 24 router hosts we requested still returned `embed.js`. Hudson Rock first reported the poisoned scripts, Report URI published the Optimizely timeline and wallets, Censys documented the MCH and Caishen consoles, and Amirul Cyber named the campaign. Write-up with router hosts, lures, hashes and wallets: https://ohey.ca/research/optimjack-iphone-exploit-router-for-rent/
NVMe/TCP & CUPS: Three Zero-Days, LLM Variant Hunting, Disclosure Bottlenecks… and What “Fixed” Means Now
From an NVMe/TCP authentication bypass to CUPS code execution and a cups-browsed root escalation, post examines what happens when LLM-assisted variant hunting meets sixty-day disclosure timelines; and why blocking a single PoC doesn't mean the security invariant is fixed.
N-able on Wednesday disclosed that unknown threat actors gained unauthorized access to one of the servers in the Mail Assure and Spam Experts Hosted Cloud, causing it to disable management interfaces and APIs. "We have taken steps to mitigate the threat and are continuing to harden our services to reduce the risk of recurrence," N-able said. "The investigation is ongoing, but at this time we
Posting this as a warning and to ask whether anyone recognizes this campaign. This concerns a third-party download source, not an accusation against the legitimate Mem Reduct project. TL;DR: An investigation of my Windows 11 computer found hidden cryptocurrency miners, cookie-extraction utilities, persistent startup tasks, and extensive Microsoft Defender exclusions. Browser download history strongly points to a Mem Reduct installer bundle obtained from memreduct\[.\]app. The installed Mem Reduct executable itself matches the official release exactly. IMPORTANT LIMITATIONS This was live-system triage, not a complete forensic image or sandbox analysis. The original installer and RuntimeHost.exe launcher were no longer available. The installation timeline is strong circumstantial evidence, but I do not have a complete parent-process chain proving the initial infection source. I also could not verify an exfiltration destination or establish exactly what data was transmitted. HOW I NOTICED IT My computer intermittently lagged, and I noticed command-window-type processes opening and closing. Startup entries labeled RuntimeHost were suspicious. Disabling Startup entries was insufficient: separate scheduled tasks remained enabled, including one configured to run every five minutes. PROBABLE DOWNLOAD SOURCE Edge download history recorded: Landing page: memreduct\[.\]app Download chain: 1.get-files\[.\]xyz/memreduct/\[REDACTED\] → pub-6a90252f6d70443d960d6ac029dbd446.r2\[.\]dev/memreduct-setup.zip These are defanged indicators. Do not visit or download from them. TIMELINE — OCTOBER 7, 2026 Times below are UTC−6: 02:28:05 — Edge downloaded memreduct-setup.zip. 02:31:05–02:31:08 — Windows Installer recorded a Visual C++ 2017 x64 runtime installation from the user's Temp directory. 02:31:17 — Installed Mem Reduct files were created. 02:37:29 onward — Hidden miner files appeared in a Windows cache subdirectory. 02:40:37–02:40:38 — Cookie-viewing tools appeared in a randomly named ProgramData directory. 06:21:01 — Earliest retained Defender detection found during this investigation. The Microsoft runtime installation is included only as timeline evidence; it does not mean the runtime itself was malicious. WHAT WAS FOUND 1. Hidden cryptocurrency miners Directories included: C:\\Users\\\[REDACTED\]\\AppData\\Local\\Microsoft\\Windows\\Caches\\9AE2CA11\\ C:\\Users\\\[REDACTED\]\\AppData\\Local\\Microsoft\\Windows\\Caches\\BB4B23F6\\ C:\\ProgramData\\Microsoft\\Windows\\Caches\\E8909865\\Content.IE5\\D7F39231\\ Subdirectories included: \- bzminer \- lolminer \- peakminer \- srbminer Executable names included: \- RuntimeBroker8A9E.exe \- RuntimeBrokerA3EB.exe \- RuntimeBrokerDB91.exe Static metadata and strings identified PeakMiner and SRBMiner components. These were not ordinary Windows RuntimeBroker executables. Other files included miner.dat, .b\\<miner>\\m.dat, .src, and .o. Some older payload files had an encrypted wrapper; .src files contained DPAPI-protected data. Attempts to decrypt the configuration with the available user/machine context and no additional entropy failed. 2. Browser-cookie utilities Files named ccv.exe and mzcv.exe identified themselves as: \- NirSoft Chrome Cookies Viewer \- NirSoft MZCookiesView They appeared in directories including: C:\\ProgramData\\335AAAEA\\ C:\\ProgramData\\49EECAC1\\ C:\\ProgramData\\5E5B1D77\\ C:\\ProgramData\\7AF900CB\\ These utilities have legitimate uses. Their presence in these directories, alongside the other artifacts and antivirus exclusions, was suspicious. Their presence does NOT by itself prove cookies were exported or transmitted. No such transmission was verified. 3. Persistence Scheduled tasks: \- Windows System Health — at logon \- Windows System Health Check — every five minutes \- Windows System Health Monitor — at boot They pointed to: C:\\ProgramData\\Microsoft\\Windows\\Caches\\4259B89A\\RuntimeHost.exe Registry startup values named WinSysCache existed under: HKCU\\Software\\Microsoft\\Windows\\CurrentVersion\\Run HKLM\\Software\\WOW6432Node\\Microsoft\\Windows\\CurrentVersion\\Run A startup shortcut also existed: C:\\Users\\\[REDACTED\]\\AppData\\Roaming\\Microsoft\\Windows\\Start Menu\\Programs\\Startup\\RuntimeHost.lnk At inspection, RuntimeHost.exe was absent and the tasks returned 0x80070002 (“file not found”). No matching launcher process was observed. 4. Antivirus exclusions There were extensive path and process exclusions, including cache payload directories, Temp directories, .NET Framework directories, and Program Files (x86). Process exclusions included miner names, RuntimeHost.exe, the renamed RuntimeBroker executables, and utilities such as MSBuild.exe, InstallUtil.exe, RegAsm.exe, and RegSvcs.exe. Some exclusions were enforced through: HKLM\\SOFTWARE\\Policies\\Microsoft\\Windows Defender\\Exclusions Those policy entries had to be addressed separately; initial removal through Defender cmdlets did not remove all effective exclusions. DEFENDER DETECTIONS Retained local logs included: \- Trojan:Win32/Suweezy \- Trojan:Win32/Commando.A!ml \- Trojan:Win32/Commandrob.A!ml Detected activity included commands attempting to create the scheduled tasks and add antivirus exclusions. These detection names are reported as recorded; I am not claiming they precisely identify every recovered payload. SHA-256 INDICATORS Recovered PeakMiner-related executable: 56c52c421266691daa9710fc3e5ea20a9a729cdaaf77cef234f56b799819412c Another recovered PeakMiner-related executable: a49907f78dc8aaa6aa10c7584e94d76a96b3637d9b730c4ea7ea8a62259b497e Recovered SRBMiner-related executable: 6eb4e0bc37d65fdf7644ab620737e14f4c7dfa9666acde43d5f38ab1e9094ac1 ccv.exe: 737b66f34e33a991a6f983b5164d800f358376a10a43cd88175259a83b986556 mzcv.exe: 0fbcaa65ada37326741259d2ebc96d52e61d38cd6c28823194f2ffb4bf906ebe These identify files recovered in this deployment. Some components are legitimate dual-use software; a matching hash alone does not establish malicious use elsewhere. THE LEGITIMATE MEM REDUCT APP My installed memreduct.exe was compared against the official Henry++ GitHub release, version 3.5.2, 64-bit. Both SHA-256 hashes matched: dd55a81d56e8c32918e5190b2ec2a1f2a7edbbca884627439b4b619fdfb5602d That supports a genuine application bundled with separate unwanted payloads, rather than a modified Mem Reduct executable. Official project: [https://github.com/henrypp/memreduct](https://github.com/henrypp/memreduct) NETWORK / DATA-THEFT FINDINGS No attacker-controlled collection endpoint was verified. The miner contains functionality for Stratum mining and optional pool telemetry, but embedded functionality is not evidence of the actual pool, wallet, or data transmitted on my computer. Historical network logs were unavailable. I am treating browser sessions as potentially exposed because of the cookie utilities, without claiming confirmed account theft. CLEANUP STATUS AT TIME OF WRITING Removed: \- Three related scheduled tasks \- Two Run entries and associated startup-approval records \- RuntimeHost startup shortcut \- Seven identified payload directories \- 39 path exclusions and 23 process exclusions, including matching policy entries The verified legitimate Mem Reduct executable and its settings were retained. Defender signatures were updated and a full scan was started. The machine has NOT yet been declared clean. Investigation records were preserved. Does anyone recognize the RuntimeHost / WinSysCache persistence, these cache-directory patterns, or the encrypted .src configuration format? I would particularly appreciate independently verified information about the loader and its network destinations.
Ruby contains seven memory-safety flaws in its regular expression engine and its string case-conversion code. For five of them, its maintainers say, proof-of-concept exploits, or PoCs, have shown that crafted input can run code on the machine. "We are aware of PoCs that prove some of them can be used to trigger remote code execution," the maintainers said. The project released fixed versions
Low-privilege database accounts that cannot read a single application table can still be used to run code on the database server itself, a researcher has shown, by reaching memory-corruption bugs in the engine code behind ordinary SQL. The exploits, published yesterday, are the researcher's proofs of concept against PostgreSQL, MySQL and ClickHouse, plus a separate Oracle chain that needs no
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Wednesday added four vulnerabilities impacting Synology DiskStation Manager (DSM), BerriAI LiteLLM, Apache Axis, and Vanderbilt University Research Electronic Data Capture (REDCap) to its Known Exploited Vulnerabilities (KEV) catalog, citing evidence of active exploitation. The vulnerabilities are listed below -
This series will describe how everyday access package configurations within entitlement management can lead to scenarios where an attacker could elevate privileges. This post lays out the fundamentals of Microsoft Entra ID Governance's entitlement management, and why an attacker may target for privilege escalation.
I tried [uncensored versions](https://openuncensored.com/models) of DeepSeek v4.1 and GLM 5.3 last night. It was quite shocking how good they were and the fact that there's nothing they won't do (like seriously, nothing). Along with that, they are cheap and run quite fast, so anyone with good enough incentive could launch swarms of them, right? How do you think the future looks? Is there any way to protect against such swarms in the future? Would it be too expensive to do so for smaller companies?
BOSSAI: a single operator’s 12-agent AI attack mesh that hit Taiwan’s emergency services (and 10 other countries)
Our Threat Research Unit tracked down a lone operator in Taiwan running a mesh of 12 AI agents - some abliterated/uncensored, some commercial - to conduct offensive campaigns across 11 countries. A few things that make this one stand out: * The operator built specialized agent roles: builders (Grok, Claude Code, Codex) handled infrastructure, while attackers (mainly an agent called Hermes, plus Cursor and others) scanned targets and ran exploits - all stitched together via 13 MCP servers wrapping tools like Kali, Ghidra, and sqlmap * Confirmed breaches hit universities and emergency-services systems, with exfiltrated medical data and PII - including a Taipei fire department’s patient-care records (ECG data, ambulance video, vitals) and a city’s 119 dispatch case records * Commercial models (Grok 4.5, MiniMax M3) consistently refused offensive requests and even lectured the operator on Computer Crime Act violations - forcing a pivot to locally-run abliterated models (DeepSeek, GLM, Qwen variants) for actual attack execution * The “offensive” models weren’t reliable either - one scanner hallucinated a critical RCE that didn’t exist, and over half of its high/critical findings turned out to be false positives * Weak OPSEC (reused branding, hardcoded API keys, a shared home IP hosting the attacker’s own legitimate side businesses) let TRU trace the whole operation back to a specific individual in Hualien City, Taiwan This case is a clear example of what’s being classified as an **“AI-augmented threat” (AAT)**, not quite a traditional APT, but a single person achieving APT-like reach across government, military, financial, and emergency-services targets by orchestrating AI agents instead of a human team. Full writeup with the MCP toolchain breakdown, model-by-model refusal/jailbreak analysis, and IOCs is available here - [https://www.acronis.com/en/tru/posts/bossai-inside-the-ai-attack-mesh-that-targeted-taiwans-emergency-services/](https://www.acronis.com/en/tru/posts/bossai-inside-the-ai-attack-mesh-that-targeted-taiwans-emergency-services/)
An updated Clairic and Clairic-flash are now open-weight, from the new Workers AI Superintelligence team
Cloudflare has become a neolab. We’re building our own frontier AI models for our customers and the open-source community. Today, we’re announcing two open-weight model releases: We’re introducing an updated version of Clairic . Since the preview, we’ve improved extraction from complex documents including images and PDFs (new!), with more reliable handling of tables, lists, and dates, and better grounding of answers in the source material. We’re introducing Clairic-flash , a lighter, faster version of Clairic that delivers strong extraction and prediction performance in a compact, 9-billion-parameter model. Both extraction models are hosted on Workers AI and full open weights are available on Hugging Face . We’re also excited to re-introduce ourselves as the Workers AI superintelligence team, the frontier AI research team behind Clef , Clairic, and lots more in the works. This team is working quickly to push the AI frontier and build useful models for Cloudflare’s customers and the open source community. A new version of Clairic: redesigned after the preview release Since announcing Clairic in preview, we have received a ton of use cases and helpful feedback on how customers wanted to use an extraction model. We heard about use cases ranging from extracting data from images like receipts to reading values from support ti
Microsoft saw an AI offensive framework move from a public GitHub secret to Key Vault data in five minutes
Microsoft’s report describes ARTEX being used for more than automated reconnaissance. The framework searched public repositories for cloud and AI-service credentials, authenticated as application identities, mapped Azure resources and retrieved additional secrets. Failed API requests were followed within seconds by corrected methods or API versions. In at least one observed case, an agent moved from finding an application secret on GitHub to extracting Azure Key Vault data in five minutes. The practical problem is less a new exploit than the speed at which a leaked credential can now be discovered, understood and used across connected services.
A pair of serious vulnerabilities in the CreditNirvana platform would have allowed attackers to access troves of detailed sensitive information, including emails, texts, and credit reports.
Talking to Sony headphones from a sandboxed macOS app: RFCOMM framing, byte escaping and a two-value sequence number
Six critical vulnerabilities in F5 BIG-IP allow an attacker to crash the Traffic Management Microkernel (TMM) without logging in, disrupting traffic until the TMM restarts. Each flaw needs a specific setup on a virtual server: an iRule that uses a particular command, or an APM policy that includes two particular kinds of agent. F5, which makes BIG-IP, disclosed the flaws today and says they
I work at Bitrise, though not on the security team. Our folks published an advisory for our recent incident on November 13 with IOCs in CSV + STIX 2.1, TLP:CLEAR, so share it wherever ([https://bitrise.io/security-advisory-november-2026](https://bitrise.io/security-advisory-november-2026)). They wrote it for security teams because the same attacker may go after other companies, so I’m posting it here. It’s still preliminary and some of it might change. Longer writeup from if you want context ([https://bitrise.io/blog/post/the-bitrise-security-incident-preliminary-review-and-recommended-steps](https://bitrise.io/blog/post/the-bitrise-security-incident-preliminary-review-and-recommended-steps)). The main thing in there is the UA family. The tooling called itself authorized pentesting or read-only research in the UA string, e.g. authorized-pentest/bitrise-readonly. Two of the variants had our company name in them, so the advisory says to expect variants with your own name if this actor shows up at your place. The regex for the whole family is in section 2.3. norma/0.4 is another one, a rare client string that showed up on all three of the main IPs. They also accessed a public app that one of our customers uses for their own pentests, which made the early scanning look like it belonged to a real engagement. To us it looked like normal pentest traffic right up until the day of the actual intrusion. A few other things the advisory suggests alerting on: * SSRF canary paths like /\_i6084\_probe\_missing and /nope-1774. These arrive from your own service’s egress, not from the attacker * sign-ups with disposable emails and “probe” in the name * \~40 different service accounts tried from one IP in 7 min * a key making 82 calls and getting PERMISSION\_DENIED on every one How they got in was a lot more boring, an internal bot account that had a valid password and no 2FA. All bot accounts were disabled in the first hours, all our production credentials got rotated, and the build infrastructure was rebuilt from scratch. I’ll pass any technical questions on to the security team, and if you get a hit (that isn’t a pentest you actually paid for), our team will get you a call with them directly.
I reverse-engineered Toy Story 2: Buzz Lightyear to the Rescue (PC, 1999) and rebuilt its engine as open-source in Rust.
A critical security flaw in Apache Roller, an open-source, Java-based blogging platform, has come under active exploitation in the wild, according to the Apache Software Foundation (ASF) The vulnerability, tracked as CVE-2026-129448 (CVSS score: 9.8), is a case of unauthenticated deserialization in the XML-RPC endpoint that could lead to remote code execution. "Deserialization of Untrusted
I did a fully static analysis of a sample of **DarkCore**, a newly emerged RAT (first advertised on forums in August, infections seen since late October according to BlackFog). This build is v3.8.0. The chain has five stages: * **VBS**: hundreds of junk lines; the real blob is split across four functions, then base64 + XOR decoded into a second VBS. * **Stage 2 VBS**: reads its *own original file* and rebuilds a payload hidden in 1,900 comment lines (base93 + a custom Feistel cipher), producing a PowerShell script. * **PowerShell**: decrypts a .NET loader and the final payload, and loads the DLL in memory, finding AppDomain.Load(byte\[\]) by its signature via reflection instead of calling it by name. * **.NET loader** (Crypto Obfuscator, DES-encrypted string tables): process hollowing into RegSvcs.exe. * **Native x64 RAT**. Some highlights from the RAT: * The config is XOR-encrypted in a custom .data2 section; I recovered the C2 (162.55.4\[.\]179:4869) and a 32-byte pre-shared key. * The C2 channel is surprisingly solid: X25519 ephemeral handshake, BLAKE2s KDF mixing in the PSK, ChaCha20-Poly1305 packets, basically WireGuard’s crypto stack. Captured traffic can’t be decrypted even with the config. * Plugins arrive as AES-GCM-encrypted PE modules. * Strings point to HVNC, remote desktop, keylogging, DPAPI credential theft and ETW/AMSI tampering. HVNC lets the operator act inside the victim’s already-authenticated browser sessions, which defeats a lot of anti-fraud checks. * Even though there isn't a geolocalization about the development of this malware, there’s a CIS geofence (Russia, Belarus, Kazakhstan, Uzbekistan, checked via GeoID, locale, time zone and keyboard layout) that makes the bot quit silently before contacting the C2. This can be an hint that the first development was made in Russia. It was disabled in this build. Full analysis can be found here: [https://inkman97.github.io/darkcore-analysis/](https://inkman97.github.io/darkcore-analysis/)
Artex and similar penetration testing tools allow newbie hackers to scan targeted systems for vulnerabilities and successfully wage attacks using prompts fed to chatbots.
On November 17, 2026, the security research company Calif disclosed Java4Shell (CVE-2026-114938) , a memory-corruption vulnerability in the Base64 decoder of OpenJDK, the most widely used Java runtime. Under the Common Vulnerability Scoring System (CVSS) 4.0, it scores 9.2 out of 10 (Critical). The flaw affects Java. OpenJDK published a security advisory and Oracle released patched builds in its November 2026 Critical Patch Update the same day, at 21:00 UTC. The vulnerability lets a remote attacker with no credentials send malformed Base64 data to a Java application and overwrite memory next to the decoded output. Calif turned that overwrite into a leak of heap memory, similar to Heartbleed , and into remote code execution. Calif demonstrated pre-authentication remote code execution against Keycloak, Atlassian Jira, Atlassian Confluence, JFrog Artifactory, Jenkins and PingFederate in their default configurations, and post-authentication remote code execution against Apache Kafka, Apache Solr, Elasticsearch, Gerrit, and Spinnaker. Cloudflare has released a new set of rules in the Cloudflare Managed Ruleset designed to detect malformed Base64 input patterns associated with this vulnerability. The rules are available on Pro, Business and Enterprise plans and are enabled by default with the Log action because malformed input and some benign application-specific token formats can resemble exploitation attempts. Blocking by default could therefore disrupt legitimate traffic. Customers can review matching requests in Security Events now and change
US secretary of state’s comments come as vulnerability in programming language Java highlighted There is “zero evidence” that a wave of IT outages – which have grounded flights, cut off work access, limited bank access, affected Apple and disrupted classified communications – are related to each other or are the work of malign actors, according to the US secretary of state. Marco Rubio said the US was “not under systemic attack” and a cascade of network problems was “not inter-related”. Continue reading...
The suspected Chinese advanced persistent threat (APT) group behind the sophisticated EggStreme malware has rebuilt its toolkit as it continues to target government organizations in the Philippines. "When attackers are exposed, the cheap response is to rotate," Bitdefender said in a technical report shared with The Hacker News. "A file hash changes with every recompile, a domain costs a few
🤖 AI is changing how threat actors operate. Attackers are leveraging AI-assisted tools to identify and exploit vulnerable systems, particularly publicly exposed services. What does this mean for defenders and the DF/IR community? Let's talk about it. 👇 [https://www.youtube.com/watch?v=X0l2ka44Ey4](https://www.youtube.com/watch?v=X0l2ka44Ey4) More at [youtube.com/13cubed](http://youtube.com/13cubed)
On AMD-based PCs and some Intel PCs, the TPM lockout for Windows Hello and BitLocker PINs can be reset by someone who opens the PC's case, researchers at Technische Universität Berlin say. That leaves short PINs open to guessing. In lab tests, the researchers found a four-digit Hello PIN on an AMD laptop and a 6-digit BitLocker PIN on an Intel laptop this way. Both PINs were set by the
TL;DR Attackers move first. On average, exploitation now starts before a patch exists (Mandiant M-Trends 2026). In 1H 2026, 23% of vulnerabilities VulnCheck added to its exploited list were hit on or before disclosure day. AI pentesting vendors aren't interchangeable. Most look convincing in a demo. The gaps show up in your logged-in flows, your coverage, and your production safety. Make
AI is speeding up how software is built, which means more code, more changes, and more chances for vulnerabilities to reach production. It's speeding up how attackers find and exploit them too, with time-to-exploit now down to a single day. Security testing has to keep up with both. AI pentesting has emerged as a way to run in-depth tests whenever teams need them, instead of waiting for the
WhiteArmor: a 3-week-old extortion crew has already hit Maersk, J.B. Hunt, Ryder and Echo, and is openly recruiting insiders for Okta/VPN/Jenkins access
We’ve been tracking WhiteArmor since its leak site appeared on 20 October. In about three weeks it has named 17 victims, with an unusual concentration in transportation and logistics: Maersk (which has confirmed a breach of an acquired legacy system), J.B. Hunt (disclosed, now facing class actions), Ryder, and Echo Global Logistics, all claimed inside a 15-day window. The more interesting artifact is a recruitment post the group calls “Project Trojan Horse.” Rather than just leaking data, it openly solicits initial-access brokers and insiders and names exactly what it wants: Okta, Azure, IAM, Jenkins, and VPN appliances (Fortinet, AnyConnect, Citrix, SonicWall). It sets a $500M revenue floor for targets and excludes the CIS region. Effectively a published targeting spec. Most of the 17 are still unverified leak-site claims (only Maersk and J.B. Hunt have public corroboration), but the access list is a useful hardening checklist regardless: identity platforms, CI/CD, and VPN gateways, plus stale credentials in acquired or legacy environments. Full write-up with the victim breakdown and demand amounts (disclosure: this is our own tracking data): [https://darkwebsonar.io/blog/dark-web-most-wanted-2026-whitearmor/](https://darkwebsonar.io/blog/dark-web-most-wanted-2026-whitearmor/) Happy to answer questions on the telemetry.
Threat actors have compromised Microsoft's agent-framework package to push a credential stealer, according to reports from Aikido, Chainguard, Endor Labs, OX Security, SafeDep, Socket, StepSecurity, and Google-owned Wiz. The malicious versions, 1.26.1 and 1.26.2, were published 33 minutes apart on November 17, 2026, and are designed to run a credential stealer on every Python start. As of
I reverse-engineered Spotify's private API and turned it into an async Python library — no API key required
At least 178 top web destinations unwittingly served a variety of malware exploits against visitors over the past 24 hours thanks to a security vulnerability involving Optimizely , a company that helps websites customize user experiences and test out new features. Experts say the incident, which saw data-stealing malicious software silently injected into sites like the BBC , CNN , Equifax , Ledger and Zoom , was exacerbated by the reality that few popular websites enforce strict rules on the sources of third-party content. The Zoom.com homepage loading the malicious JavaScript file from Optimizely. Image: blog.report-uri.com. Click to enlarge. According to a blog post from the security firm Report URI , on November 16 malicious JavaScript code was served to visitors of 178 organizations through Optimizely’s platform. Report URI researcher Scott Helme wrote that the payload delivered by the malicious Javascript was rebuilt 26 times during the day, dropping a multi-purpose malware kit that could steal payment card data, cookies and screenshots, a
Ethersphere postmortem: bee-js 13.3.0 and Swarm Desktop 0.56.2 shipped the same loader via a compromised release bot
npm pulled bee-js@13.3.0 about 75 minutes after it was published. The same release bot had also shipped Swarm Desktop v0.56.2, and that build stayed the latest GitHub release for nearly eight more hours. Its macOS build was signed and notarized under the project's own Developer ID. I found it by following the bot's identity across the org instead of stopping at the npm package. At 03:26 UTC I posted the release, the commit, the loader and a containment list in the triage thread: [https://github.com/ethersphere/bee-js/issues/1295#issuecomment-6731497299](https://github.com/ethersphere/bee-js/issues/1295#issuecomment-6731497299)
NSA issues ‘code yellow’ as NSANet, used by staff to share classified material, including top-secret intelligence, fails The US National Security Agency said on Tuesday that it had experienced “temporary degradation” during scheduled routine maintenance, after reports that its classified internal communications network had gone down. In a statement, the agency said some internal systems were affected but that “mission-critical capabilities remain fully operational” and its ability to defend national security continues. It said its infrastructure is built with “multiple layers of redundancy” to keep operations running during planned maintenance or unexpected events. Continue reading...
Last month, the Swiss non-profit foundation Quad9 announced that it will withdraw its free public DNS resolver from France. This difficult decision followed months of litigation and a French court ruling that imposed daily financial penalties on Quad9 for failing to block domains associated with alleged copyright infringement. Quad9’s departure from France is both disheartening and understandable. The court’s ruling put the organization in the impossible bind of having to choose between crippling litigation and penalties, or fundamentally altering its service in a way that would undermine the privacy and security of its users. Left with no good option, Quad9 is choosing the exit door instead. The withdrawal of a basic infrastructure service from France fundamentally harms the free and open Internet, and it should serve as a warning of the damage that will result if rightsholders and policymakers continue demanding maximal blocking across every layer of the Internet. It not only deprives French Internet users of a trustworthy and important tool; it creates a real risk of disruption to their daily lives. Quad9’s resolver is used by a wide variety of individuals and organizations in France — including government bodies, educational institutions, hospitals, and utilities — and Quad9 has no way to directly notify those individuals and institutions that their resolver will soon stop working. If the affected parties do not change their resolver before Quad9 removes its service, their Internet will fail with no explanation. Quad9 is not the first public DNS resolver to withdraw from a country following blocking orders, and it almost certainly won’t be the last. Cisco previously withdrew its OpenDNS resolver from France, Portugal, and Belgium (temporarily) under similar circumstances. And a Belgian court recently issued an
Email, flights, banks; network failures are piling up. Here’s how WIRED investigates if they’re the result of hacker mayhem. (Hint: usually not.)
Introducing ForgePtr (CVE-2026-47501): from unprivileged process (neighbor) to root on the host (landlord) for GPU clouds...
It's an NVIDIA GPU Driver LPE. (Not connected to infidelity or nation state honey-potting of tech execs) It's scored as CVSS 7.8, a “local” bug. We think it's worthwhile though considering LPEs in light of neoclouds / rentable sliced GPUs. After all, in a multi-tenanted environment, local means everyone paying to be there. In our write up we also walk through how configs can be neutered, including Secure Boot. TLDR - the bug allows attackers are to rent their way to accessing someone else’s production AI on vulnerable drivers. The bug's been in the open-source driver for 4+ years, though is now patched following our responsible vulnerability disclosure process. Big thanks to the NVIDIA PSIRT team.
I built a tool at work (Advens) for analyzing `.nessus` files after an internal or external Nessus scan. It is free and open source, and the app doesn't connect anywhere. You can inspect, search vulnerabilities, hosts and services and copy almost everything with one click (e.g. IP, IP:Port). You can also compare two scan files, check out trends (how the vulnerabilities change over time e.g. for periodic scans), and inspect a network map with possible attack paths (not the same as BloodHound, and more limited). Repo: [https://github.com/advens-gmbh/scan-insights-for-nessus](https://github.com/advens-gmbh/scan-insights-for-nessus) Windows and Linux downloads are available here: [https://github.com/advens-gmbh/scan-insights-for-nessus/releases](https://github.com/advens-gmbh/scan-insights-for-nessus/releases) Note: The Windows files aren't code-signed yet, and SmartScreen will warn when you first start the app. The releases are built by GitHub Actions from the public code, and every file can be verified with `gh attestation verify <file> --repo advens-gmbh/scan-insights-for-nessus`. Do you have ideas I should implement? Is there any feature that you would definitely want in a tool like this? (The images contain fake data)
Ryanair forces you to install their app for a boarding pass, so I reverse engineered it to get the pass without the app
The integration includes purpose-built Kibana dashboards for edge observability, detection rules that correlate perimeter and network activity with GreyNoise intelligence, and real-time enrichment that adds GreyNoise classifications to the IPs in your Elastic Security events.
Amid hundreds of reports of US banks experiencing outages, JPMorganChase says a technical “issue” with its “on-premise hardware systems” may be causing “slowness” with transactions at outside banks.
Overview A mismatch between Apple File System (APFS) filesystem and path resolution for multiple applications allows attackers to bypass security exclusions using alternative Unicode spellings. By supplying an alternative Unicode spelling or case-folded variant (such as ß versus ss ) that bypasses the application's file write checks, a local attacker can overwrite files wherever the process has write permissions, potentially leading to data corruption, configuration hijacking, or code execution. Multiple vendors have assigned CVEs and issued patches to address this flaw. Description Several archive tools, packaging systems, and runtimes enforce exclusion filters using pathname comparisons that do not match the underlying filesystem's resolution behavior. By using a byte-for-byte comparison (such as ß versus ss ), rather than full case folding, a vulnerable application can be tricked into silently overwriting protected files. This vulnerability exposes backup tools, forensic workflows, and CI/CD pipelines to unauthorized file modification. The CVEs assigned during coordination are as follows: CVE-2026-14978 HashiCorp go-slug could allow an attacker with local filesystem access to bypass .terraformignore exclusions and cause sensitive files to be included in Terraform slug uploads due to improper handling of Unicode normalization during path matching. CVE-2026-49401 Denos permission system enforces filesystem and execution restrictions by comparing the requested path against the path supplied to --deny-read, --deny-write, --deny-run, or --deny-ffi. On macOS, those comparison were performed using raw-byte pathname matching while
Whether staff are manipulated to open email links, hand over login details on phone calls, or get tricked by legitimate-seeming texts, social engineering tactics give malicious actors direct access to some of your most important data and systems. In 2026, multiple studies, surveys and industry analyst reports reiterate what most security professionals already know: people […] The post Key 2026 social engineering statistics appeared first on Heimdal Security Blog .
In October we looked at two sets of numbers from the devices Heimdal monitors. On the devices where we see installed software, 60% had a Microsoft Copilot app installed. On the devices where we log every DNS lookup, 13% contacted Copilot. Those figures come from different pools of devices, so they aren’t a conversion rate. […] The post Copilot is on 6 in 10 devices we monitor. Only a third of AI-active devices reach it, for now appeared first on Heimdal Security Blog .
Presently sponsored by: Find shadow AI fast. Origin's sensor catches every AI agent on your fleet, sanctioned or not, and shows what each one actually did. Email is just one of those utilities like power or water where you always expect it to just, well, to just be there. You sign up with a provider and use as much of it as you need, when you need it, and the idea of it being unavailable for any extended period of time is foreign. Microsoft's outage a few days ago was one of those exceptional circumstances, and whilst it absolutely rocked the whole internet, it's far from a reason not to decry using email for most of the things we currently apply it to. But it is a good time to look at where redundancy makes sense; maybe start with redundant channels for alerts about things that could hurt people !
Supermarket giant says no customer, team or supplier data compromised but recommends updating passwords as a precaution Get our breaking news email , free app or daily news podcast Australia’s second-largest retailer has been targeted in a cyber-attack, though it is still too early to tell whether any of its data has been stolen or compromised. Coles revealed the threat on Monday morning, saying it was investigating with support from the Australian Cyber Security Centre and global security experts. Continue reading...
Unlike familiar Captcha tests, this con asks you to press a series of keys. And that’s a way in for fraudsters to install malicious software You are looking for some gifts for Christmas and want to support small businesses when you shop. When you visit the site of one you’ve heard good things about you are asked to verify that you are human. Instead of clicking on a series of images, or typing letters that appear, just as you have seen before, you are asked to press a series of keys and then press enter. Continue reading...
Just launched on github: [https://github.com/Mafarricos/badbox-tv-audit](https://github.com/Mafarricos/badbox-tv-audit) To check if your android tv box (google tv, android tv) if there is malware and tries to fix it. Mostly because this new XCPE Botnet discovered. Let me know if it helped, it's optimized to be run by your AI agent, but if you understand scripts you can run it by yourself
Fully automated ransomware hasn’t arrived—yet. But seemingly unsophisticated attackers are using AI tools to wreak havoc.
We introduced Kitesurf , our stateless browser for AI agents that runs entirely on top of Workers, in August and followed up in September with WebMCP support, new APIs, and improved compatibility. On every occasion, we mentioned our commitment to open sourcing it so that developers could run their own browser engine on their own Cloudflare accounts if they want to. That promise is now real. Kitesurf is open source on GitHub under the Apache 2.0 license. You can read the code, run it locally, change it, and deploy your own instance to Cloudflare Workers. A browser built for agents Kitesurf runs entirely on Cloudflare Workers. It parses HTML, applies CSS, executes page JavaScript, and renders screenshots without a browser container. It supports the Chrome DevTools Protocol (CDP), so you can connect clients such as Puppeteer. Our main goal is to give AI agents a lightweight, efficient browser — one that can load websites, execute JavaScript, and interact with pages while using less CPU and memory than a full browser like Chromium, and scaling on demand across Cloudflare’s global Workers network. Read more about why and how we built Kitesurf in our initial announcement and how to use it in Browser Run’s developer documentation. What is inside the repository? Kitesurf builds on Cloudflare’s Developer Platform and APIs, alongside an
The company says a bad update—not a cyberattack or AI—caused its servers around the world to crash.
Spend controls have been at the top of the list of most requested features from developers building on our developer platform for quite some time now. With more than 10 million developers now building on Cloudflare, this matters more than ever. Although we’ve had usage notifications and budget alerts available to paid plans for a while, this doesn’t solve the problem completely; informing is not the same as controlling. You can now set spending thresholds and choose to have Cloudflare pause activity when those limits are reached, thus avoiding unexpected bills from unplanned usage. Today, we're introducing Spend Management in Beta, for any pay-as-you-go (PayGo) customer, starting with Workers and Durable Objects. Use Spend Management alongside AI Gateway spend limits to manage compute and AI inference costs across your application. AI Gateway already supports per-user and per-model budgets for Workers AI and other AI providers, applied to requests routed through your gateway. How it works You can start using Spend Management using the dashboard, cf CLI , or REST API . In the
Since 2021, Project Zero has followed a “90 + 30” vulnerability disclosure policy, allowing vendors up to 90 days to remediate a vulnerability plus an extra 30 days “soak time” for patch rollout. Our goal was to encourage faster patching and give vendors time to deliver patches to users. The advent of AI-assisted vulnerability discovery and exploitation has changed the threat landscape such that the vulnerability remediation timeline encouraged by our disclosure policy is clearly insufficient. LLMs have increased the speed at which attackers can find and exploit vulnerabilities, alongside the risk that multiple parties, including malicious ones, discover the same bug. A vulnerability reported today presents a much higher immediate risk to users than one reported last year. In this new environment, it is crucial that vendors move towards remediating vulnerabilities faster.
Overview A heap-based buffer overflow exists in the Tight JPEG decoding component of the TightVNC Windows Viewer version 1.3.10 and affected UltraVNC Windows Viewers. An attacker controlling a malicious VNC server can trigger memory corruption by sending a specially crafted screen update. Successful exploitation can crash the viewer and may allow arbitrary code execution, although only denial of service conditions were demonstrated during testing. Description TightVNC and UltraVNC provide remote desktop access using the Remote Framebuffer (RFB) protocol. Their viewer applications connect to VNC servers and decode screen updates for local display. Tight encoding supports JPEG compression of image data within these updates. The Windows viewer function DecompressJpegRect() allocates a heap buffer using a hardcoded maximum row width of 8,192 pixels but converts JPEG pixel data using the server-supplied rectangle width without adequately checking that the output fits within the buffer. A malicious server can supply an oversized Tight JPEG rectangle that causes the conversion loop to write beyond the allocated memory. In an allocation demonstrated during testing, which included 256 bytes of padding, the out-of-bounds write began at a rectangle width of 8,257 pixels. Exploitation requires a vulnerable viewer to establish a session with an attacker-controlled or compromised VNC server and process its crafted update. The server can advertise a sufficiently large framebuffer, so checking that the rectangle fits within the advertised framebuffer does not, by itself, prevent this overflow. Authentication or transport encryption does not prevent exploitation by a malicious server participating in the session. Impact An attacker controlling a VNC server can corrupt
GreyNoise is monitoring scanning, exploitation attempts, and successful exploitation of a suite of vulnerabilities being referred to as "Able2Shell" and "4EverAble2Shell." There are numerous vulnerabilities associated with this activity, so GreyNoise is opting to focus on the end-to-end intrusion activity against the product in order to expedite intelligence dissemination.
VU#580079: Arbitrary file read/write vulnerability in ImageMagick-based applications with default configuration
Overview Multiple arbitrary file read and write vulnerabilities have been identified in applications that use ImageMagick with insufficiently restrictive security policies. Collectively referred to as ImageTragick2 , these issues arise when ImageMagick processes attacker-controlled SVG or PostScript content using dangerous coders or delegates permitted by its default or application-supplied configuration. An attacker who can upload an image, or otherwise cause an application to process crafted image content, may be able to read or write files accessible to the ImageMagick process. In some deployments, these capabilities may be chained to achieve arbitrary code execution, compromising the confidentiality, integrity, and availability of the affected application and host. Description ImageMagick is a widely used open-source image-processing suite. Web and content-management applications, including Ruby on Rails, WordPress, MediaWiki, ownCloud Classic, and Redmine, may use ImageMagick for operations such as thumbnail generation, cropping, conversion, and transformation. ImageMagick is also commonly included in containerized AI image-processing pipelines. It may operate before inference to normalize or transform inputs, or after inference to convert and prepare generated output. In these environments, images may originate not only from direct uploads but also from URLs, automated agents, datasets, model outputs, or other upstream services. Two vulnerability variants have been identified: Coder invocation through SVG (MSL injection) ImageMagick's SVG processing can allow a crafted SVG <image> element to invoke a dangerous coder through its xlink:href value, for e