Gitea released 28.0.0 on 30 September 2026 (the project has dropped the old '1.' prefix) with 20 security fixes, then 28.1.0 on 6 October with nine more. The most serious, CVE-2026-103059 (CVSS 9.1 as reported by Cyber Security News), affects instances using Gitea's built-in SSH server: public keys were looked up with a case-insensitive SQL LIKE, so a forged case-variant RSA key whose private key the attacker controls could be matched to another user's account. Keys are now matched by fingerprint. Other 28.0.0 fixes close several ways for repository migrations, mirrors and push mirrors to reach internal hosts despite the outbound allowlist (DNS rebinding, multiple DNS answers and approved-domain IP skips, including CVE-2026-70357, CVE-2026-101027, CVE-2026-101029 and CVE-2026-89430); Gitea now sends Git network traffic through an internal proxy that enforces the rules at connect time. Gitea Actions fixes stop cancelled-then-rerun fork pull request runs from reaching self-hosted runners without approval (CVE-2026-104632, CVE-2026-104626) and stop a maintainer's event from running an untrusted contributor's workflow (CVE-2026-94205). There are also stored XSS fixes (container registry blobs, CVE-2026-103667) and a duplicate Git tree entry check that hid files from review (CVE-2026-95106). 28.1.0 adds fixes for push mirror API permission checks (CVE-2026-97208, CVE-2026-86684), repository media stored XSS (CVE-2026-96594) and an OAuth2 refresh grant that accepted access tokens (CVE-2026-101023). No exploitation is reported. Upgrading changes egress settings (set EGRESS_MODE = strict for deny-by-default) and needs Git 2.25 or later, so read the release notes first. Primary: Gitea blog; wire: Cyber Security News.
Gitea blog — Gitea 28.1.0 is released (6 Oct 2026)
Elastic posted 14 advisories (ESA-2026-185 and ESA-2026-187 to 199) on 6 October 2026. The highest-rated, ESA-2026-187 / CVE-2026-102406 (CVSS 3.1 8.8), is in Kibana's Fleet package installation: a user who has been given Fleet privileges to install custom (uploaded) integration packages, but no direct Elasticsearch admin rights, can claim a data stream identifier already used by another user or team in the same Kibana deployment. Fleet then applies the attacker's index and ingest-pipeline settings, so newly ingested data flows through infrastructure the attacker controls and stops reaching its intended destination. Elastic warns the interception can continue after the malicious package is removed, so affected infrastructure must be checked and repaired separately. It affects Kibana 8.14.0 to 8.19.21, 9.0.0 to 9.4.6 and 9.5.0 to 9.5.3, self-managed and Elastic Cloud Hosted, where delegated users can upload custom packages. Other notable fixes: ESA-2026-199 / CVE-2026-103009 (7.1), a cross-cluster search authorisation bypass that only affects clusters acting as a fulfilling cluster under Remote Cluster Security 2.0; ESA-2026-197 / CVE-2026-103007 (7.2), privilege escalation through the delegated manage_roles privilege when it uses wildcard or regex index patterns; and ESA-2026-193 / CVE-2026-102412 (6.5), where limited Fleet users could read Fleet Server host TLS private keys (Kibana 9.3 to 9.5). The rest are mostly authenticated denial-of-service bugs in Elasticsearch, plus an Elastic Endpoint crash loop on Windows hosts with Chinese, Japanese or Korean locales that can switch off Elastic Defend protection. Elastic does not report exploitation. Primary: Elastic security announcements.
SonicWall published advisory SNWLID-2026-0017 on 6 October 2026 covering four flaws in Secure Mobile Access (SMA) 1000 series appliances, physical and virtual SMA 6210, 7210 and 8200v. The worst, CVE-2026-102255 (CVSS 10.0), sits in the Appliance WorkPlace interface: an unintended alternate access path lets the appliance act as a forward proxy, so a remote attacker with no login can make it send requests on their behalf, reach internal functions and perform unauthorised operations (CWE-918 server-side request forgery and CWE-441 unintended proxy). The other three need an administrator account: CVE-2026-102256 (7.8) is an operating-system command injection that can give remote code execution, CVE-2026-102257 (7.2) is a Zip Slip path traversal in the Appliance Management Console (AMC) that can also lead to code execution, and CVE-2026-102258 (5.5) is stored cross-site scripting in AMC. SonicWall says it has no evidence any of the four are being exploited, lists no workaround, and says SSL-VPN on SonicWall firewalls and the SMA 100 series are not affected. Benoît Sevens of Anthropic reported the SSRF and command-injection bugs; Brian Mariani reported the Zip Slip through Trend Micro's Zero Day Initiative and the XSS through DigitalCanion SA. Important for anyone who patched last month: the builds that fixed the exploited September zero-days CVE-2026-83548 and CVE-2026-83549 (12.4.3-03526 and 12.5.0-02952) are the last affected builds in this advisory. Shadowserver tracks more than 400 internet-exposed SMA1000 appliances, and CISA has added 19 SonicWall flaws to its exploited list in four years, so expect attention. Primary: SonicWall PSIRT advisory; wires: BleepingComputer and Cyber Security News (7 Oct). UPDATE (9 Oct): BleepingComputer reported on 9 October that Previdian's honeypot network has seen exploitation attempts consistent with CVE-2026-102255: crafted OPTIONS requests to the WorkPlace Extraweb interface tried to reach the appliance's internal CouchDB service on 127.0.0.1:5984 and call a design document's _rewrite function using admin:admin credentials, a different technique from the July and September SMA1000 SSRF attacks. Previdian has not established whether any attempt succeeded, and SonicWall's advisory had not marked the flaw as exploited. Shadowserver tracks more than 400 SMA1000 appliances exposed online. UPDATE (13 Oct): The Hacker News reported on 12 October that the first exploitation attempts were seen by two threat intelligence firms, Defused Cyber and Previdian, on 8 October, two days after the fix shipped. Previdian has counted 135 attempts so far from two IP addresses located in the United States; those are sensor hits, not confirmed compromises, and the IP locations say nothing about where the attackers are. SonicWall had still not updated its advisory to mark the flaw as exploited. Earlier SMA1000 zero-days were used to steal credentials, session databases and TOTP multi-factor seeds before moving into internal networks, so treat an unpatched appliance as a possible foothold.
SonicWall PSIRT — SNWLID-2026-0017, SMA1000 series multiple vulnerabilities (6 Oct 2026)
Progress published a critical security bulletin on 6 October 2026 for CVE-2026-91140, an OS command injection flaw in the GenAI agent and prompt definitions for the DataDirect Autonomous REST Connector (ARC) AI Model Generator, distributed through the public progress/datadirect-arc-ai-model-gen GitHub repository as an Early Access feature. The generator takes a filename value from the OpenAPI or Swagger document it is given and uses it in a shell command for temporary-file cleanup without proper validation or quoting, so a document with shell metacharacters in that value runs commands on the developer's machine or the CI runner when someone invokes the generator. Progress rates it CVSS 3.1 9.6 (network, no privileges, user interaction required). The affected files are ARCGenAI-Generator.agent.md 2.0, ARCGenAI-Generator.prompt.md 1.0 and ARCGenAI-EntityGen.agent.md 1.0; all three are updated to 2.1. There is no installer or migration: the fix is pulling the latest definitions. Progress also tells anyone who already ran the agents on untrusted or third-party API specifications to review those workspaces and CI environments for unexpected files or commands, since the flaw produces no product error message. CISA's assessment records no known exploitation. Primary: Progress bulletin and the GitHub fix commit; wire: Cyber Security News (7 Oct).
OpenSSH 10.6 / 10.6p1 was released on 6 October 2026 with a batch of security fixes for both the ssh client and the sshd server. The headline change disables the LZ77 dictionary coder in compression: researchers Fabian Bäumer and Marcus Brinkmann showed in "Crossing the Streams" that, when compression is on, all channels in one SSH connection share one compression dictionary, so an attacker who can feed chosen text into one channel and watch ciphertext lengths can recover secrets carried in another channel. Compression still works but is less effective. The sftp client now validates paths returned by a server more strictly, so a malicious server can no longer steer a recursive download into writing outside its target folder (reported by Junghoon Cho). The ssh client now refuses destination usernames containing $ or \ on the command line, because usernames from untrusted sources could reach a shell through ProxyCommand or Match exec (reported by KeenLab Tencent); usernames set with the User directive in config files are not restricted. Other fixes cover GSSAPI credentials from a failed attempt persisting into a later login, the authorized_keys "restrict" keyword not applying to tunnel forwarding, compressed packets inflating past the size limit, and a root-privileged session process on old platforms without file-descriptor passing (GatewayPorts and StreamLocalForwarding are now forced off there). The OpenSSH team says it is getting many AI-assisted bug reports, has seen AI-found bugs rediscovered independently, and will ship releases more often. The release notes assign no CVE numbers or CVSS scores, but MITRE published CVE records for the fixes on 6 and 7 October. UPDATE 15 October: Red Hat tracked the root-retaining session-process issue as CVE-2026-110929 (Important, draft CVSS 3.1 8.1); it has since rejected that ID as a duplicate of CVE-2026-106589, which it rates Low (draft CVSS 3.1 2.9). An authenticated user on an affected build can act as root, including making privileged socket connections and binding forwards to low-numbered ports. Red Hat says Red Hat Enterprise Linux, and Linux generally, is not affected because the post-authentication process drops to the logged-in user there. The flaw applies to QNX 6, SCO OpenServer 5 and OpenSSH builds made with --disable-fd-passing, which should upgrade to 10.6. Primary: OpenSSH 10.6 release notes; wire: Cyber Security News (7 Oct). Correction (22 Oct 2026): this card gave the root-retaining session-process flaw as CVE-2026-110929 with Red Hat's draft score of 8.1 (Important) and said the 10.6 fixes had no CVEs. Red Hat has rejected CVE-2026-110929 as a duplicate of CVE-2026-106589, which it rates Low (draft CVSS 3.1 2.9), and MITRE published CVE records for the 10.6 fixes on 6 and 7 October. (source: Red Hat CVE database)
OpenSSH — Release notes, OpenSSH 10.6/10.6p1 (6 Oct 2026)
Veeam released Veeam Backup & Replication 12.3.2 P4 (build 12.3.2.4934) on 6 October 2026 and listed the security fixes in KB4934. The serious one is CVE-2025-64393, rated Critical at CVSS 4.0 9.4: a low-privileged user holding only the Backup Viewer role can run code on the Veeam Backup Server through insecure deserialization of data received by the Mount Service, with no user interaction. CVE-2026-93026 (6.1) lets a Backup Viewer user change or delete the Enterprise Manager master key and read or overwrite stored antivirus update credentials, and CVE-2025-64392 (4.8) is a reflected cross-site scripting flaw in Veeam Backup Enterprise Manager that needs a logged-in portal user to open a crafted link. These three affect build 12.3.2.4854 and earlier version 12 builds; Veeam says version 13 is not affected by them. The same patch also carries the fix for CVE-2026-58069 (8.3), first published in KB4902 for version 13.1, which lets an authenticated Veeam Cloud Connect tenant read any file on the service provider's host; for version 13 that one is fixed in 13.0.3.63 and 13.1.0.411. All four were reported through HackerOne, and no exploitation has been reported. Backup servers are a favourite ransomware target, and Veeam flaws have been weaponised quickly after past patches. Version 12 reaches end of support on 28 February 2027. Primary: Veeam KB4934.
Veeam — KB4934: Vulnerabilities Resolved in Veeam Backup & Replication 12.3.2 P4 (published 6 Oct 2026)
The SPIP maintenance team released two critical security updates on 6 October 2026 for SPIP, the French open-source content management system widely used by public bodies and associations in France. SPIP 4.4.27 (4.4.26 was skipped) fixes many flaws, which the team describes as pre-authentication remote code execution, privilege escalation, cross-site scripting, SQLite injection, anonymous HTML upload and reconstruction of the site secret. SPIP warns that its security screen, the protective filter many sites rely on between releases, does not block these. Separately, the Crayons plugin, which lets editors change content directly on the page, was updated to 3.5.0 and the Simple logs (Simplog) plugin to 2.3.0. One Crayons flaw, CVE-2026-104070 (CVSS 3.1 9.8, CVSS 4.0 9.3), lets an unauthenticated attacker skip the authorization check in crayons_store.php by leaving out the anti-forgery parameter, then chain it to write a malicious template file, read the configuration that holds the site secret, and run PHP code as the web server user. The Simplog flaw allows reading and deleting any file in the installation, which can be used to reinstall the site and create a webmaster account. No exploitation has been reported. Primary: SPIP blog; CVE record.
SPIP Blog — Mises à jour de sécurité critique : plugins Crayons et Simplog (6 Oct 2026, French)
HPE published security bulletin HPESBNW05156 on 6 October 2026 for HPE Networking AOS-Switch (AOS-S), the operating system on the former Aruba campus and access switches. It covers nine vulnerabilities in AOS-S 16.11.0031 and earlier, mostly found by HPE's own researchers. Three are rated 9.8 and need no login: CVE-2026-76742, an authentication bypass in the web management interface; CVE-2026-76743, an authentication bypass in the management interface that depends on conditions outside the attacker's control; and CVE-2026-76744, a group of buffer overflows that can let a remote attacker run code. Others include memory corruption reachable from an adjacent network that can lead to code execution (CVE-2026-76745, 9.6), adjacent and remote buffer overflows that leak information (CVE-2026-76746, 9.3; CVE-2026-76747, 9.1), privilege escalation through the API for a logged-in user (CVE-2026-76748, 8.8), and two 6.5-rated issues covering denial of service and information disclosure (CVE-2026-76741, CVE-2026-76749). HPE says it knows of no public discussion or exploit code, but urges customers to patch because of how broad and serious the set is. Versions past End of Maintenance should be treated as affected. Primary: HPE security bulletin.
HPE — HPESBNW05156 rev 1: Multiple Vulnerabilities in HPE Networking AOS-Switch (AOS-S) (6 Oct 2026)
Google moved Chrome 155 to the Stable channel on 6 October 2026, version 155.0.8059.39/.40 for Windows and Mac and 155.0.8059.39 for Linux, rolling out over the following days and weeks. The release fixes 247 security bugs. Google rates four as Critical, all use-after-free memory errors: CVE-2026-106382 in Chromecast (found by Google), CVE-2026-106197 in the browser process, and CVE-2026-106358 in Navigation and CVE-2026-106347 in Track, the last two reported by Xinyang Ge of Anthropic with help from Claude. Another 53 are rated High, among them several uninitialised-resource and type-confusion bugs in the ANGLE graphics layer, a SiteIsolation authorization flaw (CVE-2026-102322), and race conditions in Fonts; the rest are Medium and Low. Google's release note does not say any of them are being exploited, and it keeps bug details restricted until most users have updated. Google does not publish CVSS scores for Chrome bugs. Browsers built on Chromium, such as Microsoft Edge, Brave and Opera, will need their own updates. Primary: Chrome Releases blog.
Google Chrome Releases — Stable Channel Update for Desktop (6 Oct 2026, Chrome 155.0.8059.39)
A CVE record published through GitHub's CNA on 6 October 2026 covers a cross-site scripting flaw in Quasar, the Vue.js user-interface framework. Before 2.22.0, the server-side rendering (SSR) meta plugin built title, meta, link and script tags from values passed to useMeta() without HTML encoding, then added them to the raw server response. If an app puts attacker-influenced text such as a post title, product name, excerpt or display name into page metadata, an attacker can break out of the tag and inject script that runs for every visitor before the page hydrates. Client-side rendering is not affected because it sets attributes through the browser's DOM APIs. The CNA scores it 10.0 (network, no privileges, no user interaction, scope changed), though real impact depends on whether an app feeds user content into useMeta() on SSR pages. The fix shipped in quasar 2.22.0 in July 2026; the current release is 2.34.0. No exploitation has been reported. Primary: Quasar GitHub advisory and CVE record.
Quasar — GHSA-pq96-jpmf-w254: XSS via unescaped SSR meta tag rendering in getHead()
JetBrains, as CVE numbering authority, published CVE-2026-106218 on 6 October 2026: in TeamCity before 2026.1.3 and 2025.11.7, the Kotlin DSL sandbox could be escaped, leading to remote code execution on the TeamCity server. JetBrains scores it 8.8 under CVSS 3.1, needing low privileges and no user interaction, so any user able to supply Kotlin DSL project settings should be treated as able to run code on the build server. JetBrains published a second TeamCity CVE the same day, CVE-2026-106219 (CVSS 6.5): before 2026.2.1, missing validation of Git submodule URLs let users read local repositories on the server. No exploitation has been reported. Earlier TeamCity flaws disclosed in 2023 and 2024 were exploited soon after release, so internet-reachable build servers should be patched quickly. Primary: JetBrains CVE records and fixed security issues page.
JetBrains — Fixed security issues (TeamCity CVE-2026-106218 and CVE-2026-106219)
Payload, the open-source TypeScript headless CMS and application framework, had at least 29 CVEs published through GitHub's CNA on 6 October 2026 across the core package and official plugins. The worst: CVE-2026-105857 (CVSS 3.1 10.0), a crafted form submission that runs code on the server when @payloadcms/plugin-form-builder is used, fixed in 3.90.0; CVE-2026-105845 (9.8), SQL injection in the SQLite and Postgres adapters by anyone who can query readable collections with dynamic filters or joins, fixed in 3.88.0; CVE-2026-105844 (CVSS 4.0 9.3), prototype pollution through @payloadcms/plugin-import-export by an unauthenticated user that can lead to code execution, fixed in 3.88.0; CVE-2026-105863 (9.2), custom fields mapped to reserved authentication claim names placing attacker-chosen values in login tokens, fixed in 3.90.0; and CVE-2026-105858 (8.1), remote code execution through the public first-register operation on an instance with local authentication and no first user yet. Others include a second SQL injection through json or blocks fields (CVE-2026-105856, 8.6), API keys readable through ordinary document reads (CVE-2026-105849), MCP API key management outside the caller's own account in @payloadcms/plugin-mcp 3.61.0 to 3.87.x (CVE-2026-105806, 8.6), multi-tenant plugin isolation bypasses, password field access-control bypass, S3 object overwrite, Stripe proxy misuse, path traversal in local upload cleanup, and a PBKDF2 work factor below current guidance. The 4.0 canary line is fixed in 4.0.0-canary.34. No exploitation has been reported. Primary: Payload GitHub security advisories and CVE records.
Payload — GHSA-r488-j9vj-wx3q: RCE in Payload Form Builder (CVE-2026-105857)
