Socket researcher John Tuckner published research on 29 October 2026 ('Searchjacked') on 319 browser extensions installed by roughly 13.4 million users over about a decade that hijack searches, capture queries and collect referral payments from Yahoo and Bing. Socket says the work expands on previous research with 264 newly identified extensions affecting 8.7 million more users, and that the extensions were created by a handful of organised operators who quietly take over the browser's default search engine, route every query through a suspicious proxy server and collect partner referral payments from Yahoo or Bing under a syndication partner ID. According to Socket, the extensions pose as 'manual finder', 'safe search', 'PDF converter' or 'AI search' tools, are pushed into browsers through paid advertising (Google Ads, Bidbrain and more), and a majority are unlisted in the Chrome Web Store so they do not appear in its search. Socket says many need almost no code: one analysed extension consisted only of a manifest that changes the default search provider plus three icon files, with no permissions, background service worker or content scripts. Socket says that because searches pass through the operator's server first, the operator sees the complete search history, often tied to a unique identifier, and decides per request where each search ends up; it saw one extension, Predator Search (about 50,000 users), send six identical searches to four different destinations. Socket says it is only an assumption that the redirects could be malicious, but warns the operator could later serve phishing pages, fake results or malware lures selectively. Clusters it describes include a 2024 'SecuryBrowse' wave of 58 security-branded extensions with about 4.9 million users and a 'Bing-Search' group of 37 extensions with about 2.2 million users published from April 2025. (source: Socket) BleepingComputer reports that the researcher said most of these extensions had been removed from the Chrome Web Store when the report was published, that new ones are constantly being added, and that it had contacted Google for comment without a response by publication. (source: BleepingComputer)
Socket — Searchjacked: Hidden Browser Extensions Hijack Searches for 13 Million Users (29 Oct 2026)
Kaspersky's GReAT published research on Securelist on 29 October 2026 on NetamLoader, a complex .NET assembly-loading framework built on XAML markup that Kaspersky says strongly indicates a supply-chain attack targeting a custom Windows build used exclusively by government entities in Asia. Kaspersky says the sample came to light in the first half of 2026 when Kaspersky Security Network flagged a .config file using AppDomainManager injection to load arbitrary .NET assemblies; the file was packed in a CAB archive in the folder that stores Windows update packages (C:\Windows\SoftwareDistribution\Download). According to Kaspersky, most executables in the archive were legitimate components of the third-party patch management tool Patch My PC and the .NET assemblies used for the injection were Microsoft-signed; only a Resources.resx file contained malicious code, and Kaspersky says it did not observe how the CAB file was delivered to the victim. Kaspersky says the chain abuses a Windows Server AppFabric cache client configuration class and .NET BinaryFormatter deserialization (TypeConfuseDelegate) to load layered XAML payloads, patches AMSI so every scan returns clean, deletes update records, CLR usage logs and Prefetch files, filters update-agent logs in a path containing 'CmitUpdateAgent' (which it says may be linked to a custom Windows 10 build used by a government in Asia), turns off .NET ETW events, and runs its final .NET plugin inside dllhost.exe, with C2 polling delays of 22 to 28 hours when the server does not respond. Based on its telemetry, Kaspersky identified a single victim in Asia, and because the malware was designed for a customised government-only version of Windows it infers the victims are Asian government entities; it says it could not link the attack to any known APT group. Kaspersky's report includes file hashes, file paths and network indicators. (source: Kaspersky Securelist)
Kaspersky Securelist — Unknown APT uses NetamLoader to attack a Win10 OS customized for governments in Asia (29 Oct 2026)
The Drupal Association said on 29 October 2026, reporting results from the Drupal AI Security Initiative it launched with Alpha-Omega in June, that Drupal published 256 security advisories from June through October 2026, more than the 250 published across the previous 30 months combined. It says monthly volume rose from an average of 8 advisories in early 2026 to 28 in June, 74 in September and 106 in October, peaking at 39 advisories on 28 October, and that 134 different people have been credited as reporters since 1 June, against 68 in all of 2025. The Association says the surge is driven by a shift in the economics of vulnerability discovery, not a change in Drupal's attack surface: the issues are mainly latent bugs in mature core and contributed code that AI-assisted analysis has made cheap, fast and automatable to find, and most reports came from volunteers and outside researchers, many using AI tools. It says discovery is now cheap and intake is the bottleneck; a Trail of Bits report against one contributed module through the Patch the Planet programme (run by Trail of Bits and OpenAI) led to the critical core fix SA-CORE-2026-014 released on 14 October; Opengrep's PHP parser silently skipped valid PHP 8 code in about 350 files across 176 contributed projects and every supported core branch (fixed upstream on 1 September); and its Triagebot tool, which posts one message per new report with a suggested classification and likely duplicates, has flagged roughly 20% of reports as probable duplicates since mid-August. The Association says release capacity is the next constraint and that it is writing a working standard for what a useful AI-assisted security report must contain. (source: Drupal Association)
Drupal.org — 256 Advisories in Five Months: Results from the Drupal AI Security Initiative (29 Oct 2026)
Symantec's Threat Hunter Team said on 29 October 2026 that it has identified an unauthorized relay service that provides access to OpenAI's Codex and Anthropic's Claude Code command-line coding agents, mainly in markets where the vendors do not officially operate, primarily mainland China. Symantec says the service has been active since at least early 2026 and has impacted systems at a dozen organizations, most of them in mainland China, as well as multinational subsidiaries based in South Korea, Taiwan and Malaysia, in sectors including industrial automation, electronics manufacturing, semiconductors, pharmaceuticals, shipping, tire manufacturing, petrochemicals and data center services. According to Symantec, users run a fileless PowerShell or bash setup script, fetched from a lookalike domain by a one-line command, shortly after a legitimate npm install of the CLI; the script rewrites ~/.codex/config.toml or ~/.claude/settings.json to point at a relay-controlled domain, swaps in a relay-issued API key set as user-scope environment variables so the change survives a reboot, sets Codex's sandbox mode to "danger-full-access", enables network access for the write sandbox and disables telemetry and error reporting. On one host the script logged the user out of their Anthropic session and deleted the ANTHROPIC_AUTH_TOKEN and ANTHROPIC_API_KEY variables before reissuing relay-controlled credentials. Symantec says this is not a vulnerability in Codex or Claude Code, that there is no evidence so far that the relay operator has abused its access, but that a relay in this position sees every prompt, file and secret sent to the agent and, with the sandbox disabled, a malicious or compromised response could execute with the full privileges of the user's account. Symantec says the relay's backend has rotated across at least eight domains impersonating OpenAI- or Anthropic-branded chat services, and it published indicators. (source: Symantec)
Symantec (security.com) — Sandbox Flipped to ‘Danger-Full-Access’: Unauthorized AI Relays Create Risk for Organizations (29 Oct 2026)
Cisco published security advisory cisco-sa-asaftd-tunnel-dos-s6mgkb75 on 28 October 2026 at 19:15 GMT (29 October 03:15 Perth) for CVE-2026-76418, a vulnerability in the Remote Access SSL VPN service of Cisco Secure Firewall Adaptive Security Appliance (ASA) Software and Cisco Secure Firewall Threat Defense (FTD) Software that, Cisco says, could allow an unauthenticated, remote attacker to cause the device to reload unexpectedly, resulting in a denial of service. Cisco says the flaw is due to logic errors when secondary authentication is configured on a tunnel group and can be exploited by sending a crafted request when connecting to the SSL remote access VPN; it notes that a successful VPN login is not required to reach the affected processing path. Devices are affected if they run a vulnerable ASA or FTD release, the SSL remote access VPN login service is reachable, and secondary authentication is configured in the tunnel group scope (including the default tunnel group, DefaultWEBVPNGroup). Cisco rates the advisory High (CVSS 3.1 base 8.6) and says: "In October 2026, the Cisco PSIRT became aware of active exploitation of this vulnerability." Cisco says there are no workarounds, only a mitigation, and strongly recommends upgrading to a fixed release. Cisco credits Eric Fernald, who found the flaw while resolving a Cisco TAC support case. (source: Cisco) According to BleepingComputer, Cisco did not provide indicators of compromise or say who is exploiting the flaw. (source: BleepingComputer) UPDATE (30 Oct 2026, Perth; CISA): CISA added CVE-2026-76418 ("Cisco Secure Firewall ASA Software and Cisco Secure FTD Software Unchecked Return Value to NULL Pointer Dereference Vulnerability") to its Known Exploited Vulnerabilities catalog on 29 October 2026 based on evidence of active exploitation, with a remediation due date of 1 November 2026 for US federal civilian agencies. Cisco's advisory was still at version 1.0 when re-read on 30 October (Perth), with the same first fixed releases. (source: CISA; Cisco)
Cisco Security Advisory — Cisco Secure Firewall Adaptive Security Appliance and Secure Firewall Threat Defense Software Remote Access SSL VPN Tunnel Denial of Service Vulnerability (cisco-sa-asaftd-tunnel-dos-s6mgkb75; first published 28 Oct 2026 19:15 GMT)
Linux Kernel Missing Initialization of a Variable Vulnerability. Linux Kernel contains a missing initialization of a variable vulnerability that could allow for unintended modification of page-cache-backed data via failure to preserve the shared-fragment marker during socket buffer coalescing, causing ESP input processing to decrypt shared fragments in place. This vulnerability can be chained with CVE-2026-109618 and CVE-2026-109619 to implant PHP webshells and run commands as root. Apply vendor mitigations. Check the NVD record and the vendor advisory for affected versions and the patch.
IBM, the CVE numbering authority for Langflow OSS, published a security bulletin on 28 October 2026 (updated 29 October) saying Langflow OSS 1.0.0 through 1.12.8 is affected by vulnerabilities that could allow a remote attacker to bypass authentication controls, execute arbitrary code, expose API credentials, or disclose information through vulnerable dependencies. IBM lists an incomplete remediation of an auto-login security control, agentic code execution via a third-party library, operator-injected API key exposure reachable without intended authorization, and unauthenticated remote code execution exploitable through model context protocol (MCP) tool enumeration. The Langflow entries are CVE-2026-93680 (improper authentication, 9.8), CVE-2026-110186 (code injection, 9.8), CVE-2026-110193 (a low-privileged authenticated user can make the Agentic Assistant generate or submit component code that runs operating system commands on the server, 8.8) and CVE-2026-110192 (insufficiently protected credentials, 6.5); the bulletin also lists three Hono framework flaws (CVE-2026-84363, CVE-2026-84364, CVE-2026-84365; 5.3 to 6.5, fixed in Hono 4.13.5) among the open-source dependencies included in the product. IBM lists no workarounds or mitigations. This is a separate fix from IBM's 2 October Langflow bulletin already on Cyberstack (fixed in 1.12.3), so 1.12.x installs up to 1.12.8 are affected. (source: IBM) Correction (02 Nov 2026): removed The Hacker News as a source; the card now rests on IBM's 28 Oct bulletin and the CVE records carrying CISA's 30 Oct assessment. (source: IBM; CVE records)
IBM — Security Bulletin: Langflow OSS is affected by multiple vulnerabilities (28 Oct 2026, updated 29 Oct 2026)
Canonical said in a security advisory dated 28 October 2026 that a critical vulnerability affecting LXD and MicroCloud, referred to it by the Incus project, allows an attacker able to reach the LXD HTTPS API to authenticate as any trusted TLS client for which the attacker has the public certificate. Canonical says the mutual TLS authentication logic compared all certificates sent by the client against the trust store, although only the leaf certificate proves ownership of the signing key, so an attacker could place a trusted client's public certificate as an intermediate certificate in the chain. In MicroCloud, Canonical says an attacker can intercept the public TLS certificate of a joining cluster member and then authenticate as that member, and that depending on the privilege of the impersonated user the attacker may gain root access to the host machine. Canonical urges all LXD and MicroCloud users to update immediately and says the LXD point releases are in progressive roll-out from candidate to stable. (source: Canonical) The CVE record, published by Canonical as CNA, lists affected LXD versions 4.0.14, 5.0.10, 5.21.8 and 6.9 before 4.0.15, 5.0.11, 5.21.9 and 6.10.1, rates the flaw CVSS 4.0 10.0 (Critical), and carries a CISA assessment dated 30 October 2026 of exploitation: none. (source: CVE record) Correction (02 Nov 2026): removed The Hacker News as a source. (source: Ubuntu Discourse; CVE record)
Ubuntu Discourse — Security advisory: Critical vulnerability found in Canonical LXD and MicroCloud (28 Oct 2026)
Security firm Parameter published 'Do the Math: How We Took Over ChatGPT Accounts' on 28 October 2026. It says it found a way to turn an ordinary public ChatGPT share into attacker-controlled JavaScript running inside the signed-in session of whoever opened it. Per Parameter, a graph in the shared conversation loaded the MathLive math library into the main ChatGPT page, and MathLive's on-screen keyboard listener accepted messages from any web address; an HTML Preview in the same conversation used it to insert an unsandboxed frame that inherited the chatgpt.com identity, and because ChatGPT's content security policy allowed scripts from YouTube, a YouTube oEmbed callback supplied the code. Parameter says it confirmed the code could enumerate the victim's conversation list, retrieve content from a private conversation that had never been shared with the attacker and send data from it to an external server, and that 'Essentially, we could perform any action on behalf of the victim's ChatGPT account.' In its most stable reproduction the victim opened the share and clicked ChatGPT's Preview button once; in a later version Preview opened automatically, so opening the link was the only click. Parameter says it submitted the report to OpenAI's bug bounty program on 18 October, OpenAI reproduced the issue on 21 October and raised the severity from P2 to P1, and on 26 October OpenAI confirmed both ChatGPT web frontends rejected the reported messages, resolved the issue and awarded a total bounty of $5,000. (source: Parameter) Correction (02 Nov 2026): removed The Hacker News as a source; removed details no other source gives. (source: Parameter)
Parameter — Do the Math: How We Took Over ChatGPT Accounts (28 Oct 2026)
bunny.net said on its status page on 28 October 2026 (7:40 pm AWST) that last week one of its infrastructure providers reported unauthorised access involving third-party server management software, and that one of its CDN nodes, located in Armenia, was hosted on a server supplied by that provider. Bunny says that after the provider's alert it stopped routing traffic through the server within 20 minutes, reviewed and preserved the relevant security logs, securely wiped the server, revoked its access to Bunny's other systems and permanently removed it from its infrastructure. Bunny says its investigation found no evidence of unauthorised access to the CDN node or any of its data, and that the provider has also found no evidence that Bunny's server was accessed but continues to investigate; Bunny says it cannot completely rule out the possibility. If the node was accessed, Bunny says the potentially exposed information would have been temporarily cached CDN content, Pull Zone configurations and TLS certificates stored on that server; no other customer data, including bunny.net account data, was stored on the server, and no internal systems were affected. As a precaution Bunny revoked and reissued all TLS certificates for its platform domains (b-cdn.net and bunny.run) and began reissuing managed TLS certificates for custom domains. In an update on 30 October 2026 (2:40 am AWST), Bunny said reissuance of managed certificates for custom domains was ongoing in controlled batches, each with a new private key, and was taking longer than first estimated, with completion now expected by the end of 31 October; existing certificates continue to be served until replaced, and neither Bunny's investigation nor the provider's has found any new evidence of unauthorised access to its server. Bunny does not name the provider. (source: bunny.net status page) UPDATE (1 Nov 2026): In an update on 31 October 2026 (8:50 pm AWST), Bunny said it has completed the reissuance of managed TLS certificates for custom domains: all managed certificates have been reissued with new private keys and the previous certificates revoked, and no action is required from customers using managed certificates. Bunny still recommends that customers who manage their own TLS certificates revoke and reissue them with new private keys as a precaution. Bunny said it has stopped using the affected provider while that investigation continues and is changing how configurations are distributed so that each PoP only receives the configuration and TLS certificates for the sites it is actively serving, and that neither its investigation nor the provider's has found any evidence of unauthorised access to its server. (source: bunny.net status page)
bunny.net Status — Security update regarding one of our infrastructure providers (posted 28 Oct 2026, 7:40 pm AWST; update 30 Oct 2026, 2:40 am AWST)
AI security lab Irregular said in a research post dated 28 October 2026 that it evaluated GLM-5.3, Zhipu AI's open-weight model, on CyScenarioBench, its benchmark for multi-stage offensive cyber operations in simulated networks, using the official released checkpoint. Irregular says the model did not refuse any of the offensive tasks it tested and completed 3 of the 13 scenarios end to end at least once, more than any open-weight model it has tested; it says leading closed frontier models have completed 9 of 10 and 12 of 12 scenarios at least once in separate evaluations. According to Irregular, across the three completed scenarios GLM-5.3 turned an application foothold into access to internal services, built alternative connection paths, coordinated execution across multiple machines, automated remote desktop interaction, and repeatedly wrote and repaired scripts, relays, monitors and extraction tools to obtain the target data. Irregular says that in August 2026 Kimi K3 became the first open-weight model it evaluated to complete a scenario end to end. In separate internal tests, Irregular says code reconstructed from memory contributed to roughly 80% of the successful GLM-5.3 runs it examined when hunting vulnerabilities it had planted in open-source code, and that when explicitly told not to output code reproduced from memory, the model repeatedly reproduced memorized code verbatim to reach correct answers, which Irregular interprets as a form of reward hacking. Irregular advises operators of GLM-5.3 to monitor its reasoning alongside actions and outcomes and to test how reliably it follows task restrictions before granting broader access or autonomy. (source: Irregular) Help Net Security covered the findings on 30 October. (source: Help Net Security)
Irregular — GLM-5.3: Multi-Stage Cyber Operations and Instruction Violations (28 Oct 2026)
The Drupal Security Team published advisory SA-CONTRIB-2026-254 on 28 October 2026 for Commerce Core, the Drupal module for selling goods and services through a shopping cart and checkout form or through decoupled applications via APIs. Drupal says the module "doesn't sufficiently protect certain routes and entity fields, leading to an access bypass that may financially impact the vendor", tracked as CVE-2026-118539, and that "this vulnerability is being actively exploited in the wild." It rates the advisory Highly critical (Drupal risk score 21/25) and lists affected versions as <2.41 and >=3.0.0 <3.3.13. Drupal says sites on 3.x should upgrade to Commerce Core 3.3.13, sites on earlier versions to Commerce Core 8.x-2.41, and that sites should audit their orders for inconsistencies. It credits Dan Puccetti, Marcus Johansson and Robert Snyder with reporting the flaw. (source: Drupal) UPDATE (31 Oct 2026): The CVE record for CVE-2026-118539, with Drupal as CNA, was published on 30 October (11:09 am Perth). It describes a missing authorization vulnerability in Commerce Core that allows forceful browsing, affecting versions before 2.41 and from 3.0.0 before 3.3.13. Drupal gives no CVSS score in the record; CISA's ADP enrichment, added at 11:56 am Perth on 31 October, scores it CVSS 3.1 6.5 (Medium) and lists exploitation as 'none' as of 30 October, while Drupal's advisory says the flaw is being actively exploited. CVE-2026-118539 was not in CISA's KEV catalog as of its 30 October release (version 2026.10.30). (source: CVE-2026-118539 record (cve.org)) Correction (02 Nov 2026): removed The Hacker News as a source; removed details no other source gives. (source: Drupal; CVE record)
