The Hacker News | #1 Trusted Source for Cybersecurity News

Of the 37,571 Grok-built apps that security company Claroc found online, 35,578 were open to anyone without a sign-in, the researchers said yesterday. They included 24 apps built by staff at the social network X and at SpaceXSI, the company that makes the Grok AI assistant. Most of the open apps were games, landing pages, and calculators, but about 1 in 10 contained internal business or personal data. The staff apps included a confidential onboarding plan for a new sales hire, a sales territory map of 594 accounts, and an X advertising proposal to a U.S. federal agency marked confidential. SpaceXSI, formerly known as xAI, has since taken down all 24 staff apps but not other users' apps, the researchers said. Besides those staff apps, the open apps contained internal data from 30 other organizations the researchers could identify, across fields ranging from construction and energy to healthcare and defense. Of those, 26 have more than 1,000 employees, and 6 are Fortune 500 co...

We tell customers three things about securing AI agents: they need their own identities, access that fits the job they’re given, and traceability that’s separate from any human’s. That was true for us at Token as well. Then, we built an agent that broke all three by default. This helped us think through how such agents should be defended. When a browser agent acts inside a user’s authenticated session, the application may see the user but not the agent, the delegation, or the task’s intended scope. By creating an agent that breaks all the security rules, Token learned how companies should be using identity to gain control, especially as agent use scales. Standards are coming, but there are steps security teams can take today. The agent we would have flagged The agent we created is named Rehearsal . Its job is to answer questions inside the Token platform by doing rather than telling. It autonomously opens pages in the browser tab, sets filters, and fills in forms. From a secur...

The in-house Python script behind eMudhra's October key compromise has now cost the company 8 more of its intermediate certificate authorities (CAs), even though it did not weaken their keys. Each intermediate CA has its own private key, which the company uses to sign the certificates it issues. The 8 intermediate CAs compromised in October had a different problem: a single line in the script cut each of their keys to one byte, the company said this week. The script also retrieved the RSA keys of the 8 newly revoked CAs from the company's hardware security modules (HSMs). The company says it cannot show that the keys were handled as required by its own rules. It revoked the CAs late last week. The company had called those CAs unaffected. "The RSA issuers went through a different path as per our assessment of the code and are therefore not impacted," it wrote on Mozilla's bug tracker on October 19. Of the 8, 6 were for website (TLS) certificates and 2 for...

Microsoft on Tuesday released patches for a total of 1,259 vulnerabilities spanning its software portfolio, the second consecutive month where the patch volume has exceeded 1,000 flaws. These cover 726 flaws in Windows, 174 in Office, 61 in Developer Tools, 59 in SQL Server, 53 in Developer Tools, and 27 in SharePoint Server. The patches also include 165 non-Microsoft-issued CVE identifiers. More than 500 flaws relate to elevation of privilege, followed by remote code execution and information disclosure bugs. With the latest round of fixes, Microsoft has addressed over 5,100 CVEs this year alone, according to Satnam Narang, senior staff research engineer at Tenable. Of these, a Moderate-severity security flaw in Microsoft Outlook ( CVE-2026-124194 , CVSS score: 6.5) has been flagged as actively exploited by Microsoft.  "Product UI does not warn users of unsafe actions in Microsoft Office Outlook allows an unauthorized attacker to perform spoofing over a network," ...

AI provider keys that customers stored on Railway were stolen in an October breach and used to commit an estimated $740,000 in unauthorized spending, the platform-as-a-service provider said yesterday. It has promised to pay those charges itself when an AI provider will not refund them. The company builds and runs apps for its customers and stores their secrets, such as API keys and database credentials, as environment variables. The attacker decrypted variables in at most 1.85% of workspaces, about 100,000, and its database queries looked for AI provider keys. In an October 26 update, the company told customers asking about refunds for unauthorized AI spend that "Railway is unable to refund customers for spend performed with compromised API keys." It encouraged them to contact the AI providers directly and offered to escalate their requests. According to the company's incident report , the decrypted data also included linked-account tokens for 1,852 users: BYOK age...

An AI model can be given a hidden backdoor by editing fewer than 10 of its rarely used neurons, without retraining the model, Microsoft researchers have shown. When a chosen trigger appears, the model outputs an attacker's command, which an agent using that model would run with the agent's own permissions. The attacker must be able to change the model's weights. The team tested the method on three open-weight models in research published this week. Microsoft says it has not seen any threat actors use the technique. The company also published guidance on defending against it. The edited neurons are ones that rarely fire and barely affect the model's answers. They were set to fire only when a trigger string appears, either in a user's request or later in a tool's output. The researchers call the method Lazarus . Each edited neuron forces one token of the command into the model's output, so a command needs about one neuron per token. The new neuron s...

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) Tuesday added two security flaws impacting Cisco BroadWorks and Openwave Messaging Webtop Media Transcoder to its Known Exploited Vulnerabilities ( KEV ) catalog, citing evidence of active exploitation. Details of the vulnerabilities are below - CVE-2026-20320 (CVSS score: 7.5) - An improper restriction of XML external entity reference vulnerability in the Open Client Interface (OCI) XML Parser of Cisco BroadWorks that could allow an unauthenticated, remote attacker to read sensitive configuration information on an affected system with the privileges of the Cisco BroadWorks user. (Fixed in August 2026) CVE-2026-111793 (CVSS score: 4.3) - An incorrect resource transfer between spheres vulnerability in Openwave Messaging Webtop Media Transcoder that could allow an authenticated user to process and convert unintended document types if bypassing the Webmail experience. (Fixed in October 2026) The addition of C...

A flaw in Git can let a malicious repository run an attacker's code on a computer that clones it recursively, together with its submodules. The Git project fixed it this week in Git 2.56.1, with matching fixes for the 2.50 through 2.55 series in 2.50.2, 2.51.3, 2.52.1, 2.53.1, 2.54.1 and 2.55.1. The flaw, tracked as CVE-2026-111760 , is an integer truncation in how Git handles its index. An extremely long path in a repository can pass Git's safety check, but Git then writes it to disk under a shorter name that an attacker can set to .git. This lets an attacker replace a submodule's .git file with a symbolic link to a folder holding malicious hooks or configuration. Hooks are programs Git runs at set points, so once Git uses that folder, the attacker's code can run. Submodules matter here because Git refuses to replace a .git folder with a symbolic link, but in a freshly cloned submodule, .git is a file. The fixed versions, listed in the Git pr...

An attacker exploited a remote code execution flaw in SGLang , an open-source inference engine, and ran code on one of NEAR AI's inference servers last Sunday. The attacker could also have seen requests being processed on that server that day, including requests sent in confidential mode, the company said earlier this week. That mode processes requests inside a trusted execution environment (TEE) designed to keep them private even from the company itself. But the TEE "does not protect against a vulnerability in the software running inside the environment," the company said. SGLang runs inside the TEE, which is how the attacker could reach requests there. That access has since been closed, but not before the attacker moved further into the company's systems. The attacker reached an internal system behind Private Chat, a product the company had already deprecated. So far, the company has confirmed that the attacker downloaded account details for about 100 users...

French software company Vates VMS has warned of a critical security vulnerability in Xen Orchestra that it said is being actively exploited in the wild. The authentication bypass flaw, which does not yet have a CVE identifier, is tracked as VSA-2026-062 . "REST cookie parsing turns authenticationToken=j:%7B%7D into {}," Vates VMS said in an alert. "REST authentication passes that object to the XO token provider without requiring a string." "The token collection treats {} as an unfiltered lookup and returns its first token; the provider authenticates as that token's owner if it is unexpired. The authenticated /rest/v0/users/me/authentication_tokens route then returns reusable login token IDs." In other words, the vulnerability can allow an attacker to bypass authentication by sending a specially crafted cookie and perform arbitrary actions with "Admin" privileges, ultimately resulting in arbitrary code execution on managed pools and ...

Three distinct threat activity clusters with shared infrastructure and malware code similarities have been observed targeting telecommunications providers, government organizations, military entities, research institutions, and critical infrastructure across Latin America, Asia, and Europe over the past year. Check Point Research is tracking the clusters under the names SilentDelivery , OceanGate , and DarkWire . Although initially assumed to be independent sets of attacks, each with its own backdoors and post-exploitation tooling, they have all been linked through a malware ecosystem called Sparrow. The entity behind the Sparrow framework has been codenamed CopperDragon , and is assessed to be a digital quartermaster developing tools for China-nexus threat actors, including FamousSparrow , Earth Estries , Tropic Trooper , UAT-9244 , and UAT-8302 . "Each cluster employed a different toolset and different initial access techniques, including compromising internet-facing inf...

Cybersecurity researchers have disclosed details of an attack that exploited two now-patched vulnerabilities in AhsayCBS to deliver a Python-based implant. The activity, observed by Huntress in early October 2026, involves the exploitation of a chain comprising CVE-2026-105133 and CVE-2026-105134 to bypass authentication and execute arbitrary commands on affected systems. Patches for the two vulnerabilities were released by Ahsay Backup in mid-October 2026. The two vulnerabilities are assessed to have been weaponized as zero-days, with the managed security platform first observing threat actors employing a smash-and-grab approach to deploy web shells and XMRig cryptocurrency miners on compromised systems. "However, this incident reflects a much larger intrusion with more patience; the operator planted shells in several places, ran two different miners, staged tunneling tools, and installed a custom implant, all on servers whose job is to hold backups," Huntress rese...

The threat actor known as Blind Eagle has resurfaced with a new campaign targeting organizations in Latin America through judicial and administrative-themed phishing lures to deliver a remote access trojan (RAT) that shares its lineage with AsyncRAT and DCRat. "The campaign uses Portuguese and Spanish lures that resemble court notifications, fiscalía notices, tutela proceedings , and administrative documents," Zenox said in a report shared with The Hacker News. Assessed to be a Spanish-speaking group and active since 2018, Blind Eagle (aka APT-C-36) has a track record of using spear-phishing lures to distribute commodity RATs like AsyncRAT, BitRAT, Lime RAT, NjRAT, Quasar RAT, and Remcos RAT. Attack chains begin with phishing emails targeting individuals who handle cases, contracts, judicial, or institutional service requests to trick them into opening a malware-laced attachment. One email containing a Colombian judicial-themed XHTML file has been identified as ...

Fireworks AI disclosed on Monday that its investigation into a security incident first reported early last month  uncovered further malicious activity, affecting account and billing information, alongside secrets, datasets, and incomplete models for a "limited set of accounts." In a notice published on October 10, 2026, the company said it detected unauthorized use of internal Fireworks credentials that allowed an unauthorized third-party to access environment variables used in evaluation jobs in a limited number of customer accounts. The activity was detected on October 6, 2026, at 9 a.m. PDT. At that time, the AI infrastructure provider said it only found evidence that environment variables, including API keys and tokens stored as secrets, were captured in certain affected accounts, adding that there were no signs that customer models, datasets, training data, inference traffic, or personal information were accessed. In response to the incident, Fireworks AI took th...

An iOS sandbox escape published on GitHub lets an app read other apps' files on iPhones running iOS 26 and 27. That puts seed phrases, private keys, and login credentials at risk of exposure to malicious apps, Binance warned today. Apple's updates for iOS 26 and 27, released yesterday, do not list a fix. A researcher who goes by forcequit ,  in collaboration with 0xjohnny ,   released the escape, called worse_query , on Sunday, a day before those updates came out. Neither the exchange nor the researcher reported any attacks or thefts involving it. The exchange told users to delete apps installed from untrusted sources, such as sideloaded or manually signed apps, and to update iOS and all their apps. No fix for the escape has been confirmed. For anyone who suspects a breach, the exchange's  security warning  advises moving funds to a new wallet created on a clean device and resetting all active login sessions. The researcher's  proof-of-concept cod...

At some point your security will fail. This isn’t pessimism. It’s not because your team and tools aren’t sophisticated enough. It’s the reason we have detections in the first place. And, it’s why once we accept that prevention has its limits, we arrive at an interdiction strategy: finding and disrupting what got in. After the alert The alert tells you something happened. However, gaining access to your network doesn’t necessarily mean attackers have stolen data or disrupted the systems they were targeting. They may need to obtain additional access, move to another machine, or further explore the network. Being able to see and understand that activity is the intervention opportunity. People naturally gravitate toward detections during product evaluations. They’re tangible, easy to compare and quantify, and address the question everyone wants an answer to: Will they catch the attack? There’s often an assumption that the rest will follow, meaning, if the detections are good, surely...