The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added five security flaws impacting JFrog Artifactory, ConnectWise ScreenConnect, and MikroTik RouterOS to its Known Exploited Vulnerabilities (KEV) catalog, following reports of active exploitation in the wild. Details of the vulnerabilities are as follows - CVE-2026-42016 (CVSS score: 8.1) - An incorrect authorization
Cybersecurity News and Vulnerability Aggregator
Cybersecurity news aggregator
Updated: 6:40 pm
J/K Navigate / Search S Save V Open Link
— or
treemd <(curl -sL https://allsec.sh/md) (as Markdown) Top Cybersecurity Stories Today
VU#369611: ExLlamaV3 contains Denial of Service vulnerability via insufficient bounds checking on kernel dispatch index
Overview An out-of-bounds (OOB) memory access vulnerability involving unchecked array indexing has been identified in the exllamav3_ext compute unified device architecture ( CUDA ) extension. Successful exploitation can lead to an immediate denial of service or application instability. This vulnerability is tracked as CVE-2026-84286. Description An OOB memory access vulnerability exists in the exllamav3_ext module due to insufficient input validation. When the kernel parameter K is set to 0 in a crafted input, the extension generates a negative array index, resulting in a CUDA illegal memory access.s. The root cause is a missing bounds check in the kernel-table dispatch process. The checkpoint-derived block index ( cbi ) is used to access a fixed 24-entry array without confirming that either K or cbi fall within safe limits. Impact Primary impacts include Denial of Service (DoS) through process crashes and potential unstable execution states within applications utilizing the library. Solution The vendor has addressed this vulnerability in the main repository. Users are advised to update their installations or apply the fix from the merged pull request: https://github.com/turboderp-org/exllamav3/pull/310. Supply chain Downstream projects utilizing this library are indirectly exposed to this vulnerability. According to the ExLlamaV3 Dependency Graph, there are 49 total dependencies currently tracking this repository. Notable downstream projects directly affected by this
AI keeps showing up in the wrong places. Attackers are using it to speed up exploits, test defenses, and automate more of the job. Some models are also crossing lines on their own. That is not a great combination. The rest of the week is more familiar: old bugs still working, fresh exploit chains, exposed systems, weak defaults, and simple paths that should have been harder to abuse. A few of
The City Attorney’s Office has asked Meta to explain how the harmful ads repeatedly ran on Facebook and Instagram. The company claims the ads are not under the city’s jurisdiction.
Before you sign with a penetration testing vendor, ask precise questions that turn marketing claims into measurable commitments. Confirm exact scope, learn which work is automated, AI-led, or human-led, and require that findings get reproduced and validated before they reach a report. Ask who can access your environment, how testers are vetted, and what the rules of engagement and stop conditions look like. Review sample reports, confirm remediation and retesting terms, and normalize every cost. Place every material promise in the contract rather than a slide deck. The post Questions to Ask a Penetration Testing Vendor Before You Sign appeared first on Synack .
Latest
Monday, September 14
Researchers have disclosed a new hardware attack, called DDRop, that breaks the memory protection in Intel and AMD confidential computing by silently dropping writes to a server's memory, so the processor keeps reading old encrypted data as if it were current. The attack requires an attacker who already controls the server's software and can briefly access the machine to insert a small circuit
A Chinese threat actor tracked as Red Heron has been attributed to the rapid exploitation of a recently disclosed security vulnerability in Gitea to compromise internet-facing instances as part of a multi-national campaign. "Red Heron scanned 1,386 Gitea instances across seven countries and maintained a separate dataset of 477 Taiwan-based systems," Acronis Threat Research Unit (TRU) said in an
In the biggest-ever legal action against harmful deepfake websites, the Manhattan District Attorney’s Office has seized 12 sites that collectively targeted around 1,200 victims.
WordPress has announced it's launching an automated security review for every release of a plugin before it's distributed through the WordPress.org update API so as to analyze it for potential security issues and ensure there are no risks involved. "New plugins are reviewed before they enter the directory, but updates ship continuously after that," David Perez, WordPress Official Plugin
AI keeps showing up in the wrong places. Attackers are using it to speed up exploits, test defenses, and automate more of the job. Some models are also crossing lines on their own. That is not a great combination. The rest of the week is more familiar: old bugs still working, fresh exploit chains, exposed systems, weak defaults, and simple paths that should have been harder to abuse. A few of
There's a lot of noise around AI and cybersecurity right now. What’s actually important is far simpler, if often lost in the hubbub. Vulnerability discovery is getting faster and happening at a much greater scale, while defenders still have to work out which findings actually deserve their action. In the first half of 2026, a whopping 35,853 CVEs were published, roughly 49% more than in the
An analysis of 160 deepfake websites reveals politicians in 22 countries appear on them. Nearly all of them are women.
A malicious cross-store Twitch browser extension has leaked OAuth tokens associated with nearly 31,000 users to proxy servers operated by a Russian commercial bot service. The extension, named "Twitch Enhanced Viewer | JeetBot," lists HISHIMIRO/jeetbot.cc as its developer and has the following identifiers on the Google Chrome Web Store and Mozilla Firefox Add-Ons store - Chrome -
Sunday, September 13
Microsoft has disclosed details of two campaigns in which threat actors are abusing third-party email delivery infrastructure to blast financial fraud scam messages and using passkey-themed social engineering to breach cloud environments. The first campaign, per the tech giant, involved sending over a million scam emails between August 3 and 5, 2026, by masquerading as chief executive officers
Saturday, September 12
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added five security flaws impacting JFrog Artifactory, ConnectWise ScreenConnect, and MikroTik RouterOS to its Known Exploited Vulnerabilities (KEV) catalog, following reports of active exploitation in the wild. Details of the vulnerabilities are as follows - CVE-2026-42016 (CVSS score: 8.1) - An incorrect authorization
Plus: The US disrupts the internet’s biggest black market, a Conti ransomware hacker gets prison time, Meta fails to stop AI-generated videos of child abuse.
Over the past year, we watched a new class of alert appear in enterprise security operations centers and grow faster than anything else in the stream: alerts that were triggered by AI tools and agents. Not attacks against AI, but the ordinary, everyday footprint of an organization using it, from developers running coding agents and non-technical staff signing consumer AI tools into corporate
The "major malicious attack" that targeted RubyGems in May 2026 was the work of a swarm of OpenAI agents, according to a new report published by researchers Spencer Kitts, Thomas Larsen, and Sydney Von Arx. On May 12, Maciej Mensfeld, senior product manager for software supply chain security at Mend.io, disclosed details of a coordinated cyber attack that targeted the package manager for the
Friday, September 11
Presently sponsored by: Report URI: Guarding you from rogue JavaScript! Don’t get pwned; get real-time alerts & prevent breaches #SecureYourSite I think what really resonates with me this week is being able to completely turn the tables on perceptions around things like AI being the big bad hacking tool the news would have you believe. There's the stat I talk about where it's had literally 0% impact on insurance payouts, the Dutch bloke who's now being tracked down based on his voice and Scott's MAG breach teardown that ultimately pointed to exposed keys rather than anything sophisticated. Then the mention in there about the Aussie headline where ChatGPT has hacked a court system down here when, in fact, a bloke used it to write a scraper he then pointed at the aforementioned website. It seems the truth just doesn't make for good headlines 路♂️
VU#369611: ExLlamaV3 contains Denial of Service vulnerability via insufficient bounds checking on kernel dispatch index
Overview An out-of-bounds (OOB) memory access vulnerability involving unchecked array indexing has been identified in the exllamav3_ext compute unified device architecture ( CUDA ) extension. Successful exploitation can lead to an immediate denial of service or application instability. This vulnerability is tracked as CVE-2026-84286. Description An OOB memory access vulnerability exists in the exllamav3_ext module due to insufficient input validation. When the kernel parameter K is set to 0 in a crafted input, the extension generates a negative array index, resulting in a CUDA illegal memory access.s. The root cause is a missing bounds check in the kernel-table dispatch process. The checkpoint-derived block index ( cbi ) is used to access a fixed 24-entry array without confirming that either K or cbi fall within safe limits. Impact Primary impacts include Denial of Service (DoS) through process crashes and potential unstable execution states within applications utilizing the library. Solution The vendor has addressed this vulnerability in the main repository. Users are advised to update their installations or apply the fix from the merged pull request: https://github.com/turboderp-org/exllamav3/pull/310. Supply chain Downstream projects utilizing this library are indirectly exposed to this vulnerability. According to the ExLlamaV3 Dependency Graph, there are 49 total dependencies currently tracking this repository. Notable downstream projects directly affected by this
Synack's reporting experience now scopes to individual assets, generates AI-written executive summaries, and lets teams preview, save and share reports without leaving the platform. The post Executive-Ready Reporting Is Here: What’s New in the Synack Platform appeared first on Synack .
The proposed class action alleges Meta illegally harvested people’s Facebook and Instagram photos to train its AI image-generation models and to build its unreleased “NameTag” face recognition feature.
GitLab has released patches to address multiple flaws, including a maximum-severity security vulnerability that has witnessed in-the-wild probes within hours of public disclosure. The vulnerability in question is CVE-2026-85706 (CVSS score: 10.0), a path traversal issue in the repository commits API that could allow an unauthenticated user to read arbitrary files from the GitLab server under
Anthropic on Thursday said it identified and disrupted industrial-scale illicit distillation attacks against Claude from seven labs based in China, including Alibaba, Moonshot, DeepSeek, Z.ai (aka Zhipu), and MiniMax. Knowledge distillation by itself is a legitimate training method. It refers to a machine learning technique where a large, powerful AI model assumes the role of a "teacher" to
Anthropic has warned that cybercriminals and state-sponsored hackers alike are using its Claude models for cyber attacks, weapons design, propaganda, and mass surveillance between December 2025 and August 2026. The threat actors, which the artificial intelligence (AI) company has branded Generative Threat Groups (GTGs), span state-sponsored groups, financially motivated criminals, commercial
Anthropic on Thursday revealed it disrupted a campaign mounted by a Russian state-sponsored threat actor that abused Claude for developing an AI-assisted workflow to get ahead of the detection curve. The operation has been attributed to a cyber espionage group it calls GTG-20006 (where "GTG" stands for Generative Threat Group), which aligns with broader reporting linking the cluster to Midnight
Security teams have become exceptionally talented at finding vulnerabilities. Now, it’s time to turn our attention to optimizing the process for determining which of those vulnerabilities actually create a path to compromise. A critical vulnerability may look alarming on a scanner report, but if it sits behind strong segmentation, identity controls, and other defenses that prevent an attacker
The new joint operation uses laser-based technology capable of detecting, tracking, and disabling commercial drones linked to human and drug trafficking.
Thursday, September 10
A lot of this week’s security news has the same awkward answer to one question: “Why was that allowed to work?” An extension asks for access and takes too much. A trusted service becomes part of a phishing chain. An old bug still gets results. An exposed system stays exposed. A package looks useful right up until it isn’t. Different stories, same basic problem: the path in was often already
VU#687587: AOMEI Backupper amwrtdrv.sys local privilege escalation vulnerability allows arbitrary writes to physical disks
Overview An incorrect permissions assignment vulnerability in the amwrtdrv.sys kernel driver, included with AOMEI Backupper 8.4.0, allows an unprivileged local user to perform arbitrary writes to the physical disk. When Secure Boot is disabled, this can be leveraged to execute arbitrary UEFI-level code before the operating system loads. This allows an attacker to bypass OS-level security controls, including HVCI, EDR solutions, and Microsoft Defender. The attack may also enable capture of BitLocker Volume Master Key (VMK) material, depending on the system's BitLocker configuration. Description AOMEI Backupper from AOMEI International Network Limited is designed to provide backup and disaster recovery services. It also helps individuals and businesses to create system images, disk clones, and file backups. AOMEI Backupper is available as a Windows application and can be integrated into enterprise backup workflows or directly used by end users. CVE-2026-12780 : An Incorrect Permission Assignment for Critical Resource (CWE-732) vulnerability in the amwrtdrv.sys kernel driver used by AOMEI Backupper 8.4.0 allows an unprivileged local attacker to achieve UEFI-level arbitrary code execution by directly writing to physical disk devices. The driver creates a world-accessible device object without a security descriptor, therefore allowing any user-mode process to open the device and issue unrestricted write requests. Hence, an attacker can modify disk sectors in the pre-partition gap (LBA 34–2047), inject a malicious UEFI payload, and alter the GPT to reference the payload as an EFI System Partition. The payload can then execute during the UEFI Boot Device Selection (BDS) phase, before operating system security mechanisms are loaded. Impact
US government adviser Paul Christiano warns of risks to AI industry as he joins OpenAI’s non-profit foundation OpenAI is not on track to reduce the risk of “catastrophic” loss of control to an acceptable level, a member of its non-profit board has said, amid spreading public and political concern that super-advanced AIs could one day wipe out humanity. Paul Christiano, a US government technology adviser, said : “There is a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic and irreversible loss of control in the very near term.” Continue reading...
InquiryIQ, a previously unreported prototype, tested a model from xAI, maker of Grok, to surface associates, social accounts, and other information about people identified through Clearview.
Wednesday, September 9
The City Attorney’s Office has asked Meta to explain how the harmful ads repeatedly ran on Facebook and Instagram. The company claims the ads are not under the city’s jurisdiction.
The new Apple Watch includes several “intelligent” listening features that have privacy and security baked in. But the protections can’t change the facts of what the tools do.
Before you sign with a penetration testing vendor, ask precise questions that turn marketing claims into measurable commitments. Confirm exact scope, learn which work is automated, AI-led, or human-led, and require that findings get reproduced and validated before they reach a report. Ask who can access your environment, how testers are vetted, and what the rules of engagement and stop conditions look like. Review sample reports, confirm remediation and retesting terms, and normalize every cost. Place every material promise in the contract rather than a slide deck. The post Questions to Ask a Penetration Testing Vendor Before You Sign appeared first on Synack .
A new GTA mod lets you smash and shoot Flock’s automatic license plate readers around the fictional Los Santos.
Fermat famously claimed to have a “truly marvelous proof” of his Last Theorem , but he never wrote it down, insisting the margin of his page was too narrow to contain it. A few centuries later, Anthropic announced a complete formalization of Fermat’s Last Theorem using 13 million lines of Lean code (clearly not what Fermat intended). Luckily, we found a wonderfully cursed Lean bug , shown below, that suggests the proof may have fit the margin after all. The issue affects all stable versions of Lean up to 4.33.1, and the patch is incorporated in v4.34.0-rc1. A “checked” proof of Fermat’s Last Theorem using Lean 4.33.1 The blue checkmarks in the screenshot above would suggest that Lean considers this proof correct. This seems odd given the amount of work Sir Andrew Wiles put into this problem and the vast size of Claude’s proof. So what is going on? The “proof” clearly doesn’t make any sense and exploits an issue in Lean. We found the issue while using GPT-5.6 to experiment with a new skill for code review. We want to clarify up front that the issue is not a kernel soundness issue , but it happens to nicely fit any discussion of strings, lengths, and substrings. &
I’ve spent enough time around AI adoption now to notice a pattern. Ask a business how AI is going for them and the honest answer is, lately, “we’ve got Copilot, and it’s not great.” That gap between the hype and the reality is exactly where Samantha North operates. Sam runs an AI transformation agency, helping […] The post AI adoption that pays off is built around keeping humans in the driver’s seat appeared first on Heimdal Security Blog .
Tuesday, September 8
Microsoft Corp. today issued updates to plug at least 974 security holes in its Windows operating systems and other software, by far its biggest single patch batch ever. Microsoft says artificial intelligence is helping to speed the discovery of vulnerabilities, but security experts warn that many organizations already are struggling to prioritize the more human-intensive endeavor of testing and deploying so many fixes each month. Image: Shutterstock.com, Kirill Makarov. This month’s patch bundle obliterates the software giant’s previous record set in July , when it released updates for at least 570 security vulnerabilities. September’s Patch Tuesday brings this year’s total to more than 2,600, more than twice Microsoft’s previous record-setting patch year in 2020 (1,245) and with three more months to go. There are two “zero-day” flaws fixed this month that are being actively exploited: both CVE-2026-81963 and
Designed to compete with OpenClaw and Instinct, the company says Muse can do everything from sell your car to book you a plane ticket.
Overview The UEFI Shell program may expose raw memory access capabilities that, if present in platform firmware for debugging or advanced support use cases, could be abused to undermine UEFI Secure Boot protections. When the UEFI Shell is included in SPI flash, an attacker with the ability to modify UEFI boot configuration may be able to create multiple boot option entries and bypass controls intended to prevent the UEFI Shell from launching while Secure Boot is enabled. This could allow an attacker to modify the pre-boot environment and execute unauthorized software during system startup. Description The Unified Extensible Firmware Interface (UEFI) is a firmware specification that defines the interface between a computing platform's hardware and operating system (OS) during the early boot process before the operating system is loaded. UEFI Secure Boot helps ensure that only trusted and digitally signed software is executed during these early stages of platform initialization. The TianoCore EDK II project provides an open-source reference implementation of the UEFI and Platform Initialization (PI) specifications. The project includes the UEFI Shell , which provides command-line utilities for debugging, diagnostics, and advanced platform management. Many OEM and Independent BIOS Vendor (IBV) firmware implementations include the UEFI Shell in SPI flash for service and support purposes. Because the shell executes in the pre-boot environment, it provides powerful commands such as dmem (display memory) and mm (memory modify) that can access physical memory. Many implementations include a boot entry for the
“You’ve been gifted a voucher!”. “Your account will be closed unless you act now”. “Late payment demand. Invoice # 207”. Phishing scams come in many shapes and sizes. And, in 2026, they remain, by far, the most common attack vector targeting both individuals and organizations. To help you understand the scale of the threat, I’ve […] The post Phishing in 2026. Latest statistics and analysis appeared first on Heimdal Security Blog .
Many security bugs are race conditions, where multi-threaded execution has to occur with the right interleaving for a negative effect to appear. This creates challenges for several use cases: Confirming bug candidates that have been discovered manually or through static analysis. Regression tests: After fixing a race condition bug, there is often no good way to write a regression test that reliably triggers the bug as part of a test suite. Automatic bug discovery, such as fuzzing: It is hard for a fuzzer to exercise all interesting interleavings of concurrent operations, or reach code paths that are only exercised when operations are racing.