Vulnerabilities (3)

A maximum-severity GitLab vulnerability (CVE-2026-85706, CVSS 10.0 Critical) was being actively exploited within a day of its September 10 disclosure. The flaw lets unauthenticated attackers read arbitrary files from self-managed GitLab CE/EE servers — including credentials, CI/CD secrets, and SSH configurations — on any instance hosting at least one public project.

watchTowr saw probing on September 11 escalate the same day to full file exfiltration, and public proof-of-concept code is now circulating. CISA added the flaw to its Known Exploited Vulnerabilities catalog on September 11, giving federal agencies until September 14 to patch.

Self-managed instances on 19.1 through 19.3 should update to 19.1.8, 19.2.6, or 19.3.2; anyone on 18.7 through 19.0 should check GitLab's advisory for their branch. If patching isn't possible, remove all public project access now, then review repository commits API logs for unauthenticated requests.

Source: Dark Reading

Patrick Wardle, founder of Objective-See, disclosed a zero-day in Meta's Muse, a macOS AI assistant Meta markets around a dedicated secure VM, protected credential storage, and user-controlled permissions. An undocumented preference called endo_voyager_dictation_endpoint can be rewritten by any process running as the logged-in user, redirecting Muse's dictation traffic to an attacker-controlled server.

From there an attacker reads dictated prompts and audio, injects instructions Muse trusts and acts on, and lifts the Muse session token. Wardle's proof-of-concept, not-a-mused, used a stolen token to make the Muse app on his iPhone report precise location, run Bluetooth scans, and list smart-home commands. No CVE or CVSS has been assigned, and no in-the-wild exploitation has been reported.

Exploitation needs code execution as the user first, so existing malware is the delivery route rather than a remote attack. Meta shipped a hot-fix roughly 16 hours after Wardle went public on September 21, and he confirmed it the next day. If you ran Muse before the fix, update it, revoke unnecessary permissions, and treat connected accounts as exposed.

Source: Cybersecurity News

Cisco has patched CVE-2026-76460, an authentication bypass in Identity Services Engine rated CVSS 10.0 that attackers were already exploiting. Insufficient authentication control on an API endpoint lets a remote attacker send a crafted request, reach the web management interface, and run commands as root with no credentials and no user interaction. Cisco ISE and ISE-PIC are both affected, regardless of configuration.

CISA added the flaw to its Known Exploited Vulnerabilities catalog on September 16 and gave federal agencies until September 19 to remediate, a deadline that has now passed. Fixed builds differ by branch: Patch 12 for 3.1, Patch 11 for 3.2, Patch 12 for 3.3, Patch 7 for 3.4, and Patch 4 for 3.5. Version 3.0 is unsupported and receives no fix.

A compromised ISE can impersonate devices, disable access controls, and open the rest of the network. Cisco says no workaround addresses the flaw, though infrastructure access control lists restricting management and control plane traffic limit remote exploitation until you patch. Check the ISE access.log for suspicious usernames, and treat a clean result carefully — root access lets an attacker remove the traces.

Source: Dark Reading

A newly disclosed flaw in Steam's Windows Client Service lets a standard local user escalate to full NT AUTHORITY\SYSTEM privileges with no admin credentials, no UAC prompt, and no game running. Exploitation needs code execution as an ordinary user and a running Steam client, which counts even when Steam sits idle at the login screen.

Researcher KillaBoi published a proof-of-concept called BrokenPipe on September 14, targeting steamservice.exe. Steam's service accepts a caller-controlled installation root that Valve's signed install script does not cover, so an attacker relocates a launcher to an unprotected path and has the privileged service execute it. No signature is forged or modified.

KillaBoi reportedly notified Valve in March 2026 and says the HackerOne report was marked a duplicate, which prompted the public release. No CVE, CVSS score, or Valve advisory exists, and Valve has not responded to press enquiries. The exploit was validated against Steam 10.96.30.42 on 64-bit Windows 10 and 11.

With no patch available, mitigation is all you have. Inventory Steam installations, remove the client where it isn't needed, and alert on unusual steamservice.exe child processes and on executables running as SYSTEM from user-writable directories.

Source: Cybersecurity News

CISA added CVE-2026-84869, a ConnectWise ScreenConnect flaw rated CVSS 9.9 (Critical), to its Known Exploited Vulnerabilities catalog on September 11, 2026. Missing authorization and improper privilege management let an attacker transfer files to a device and execute them during an active remote session, with no host confirmation.

Huntress observed exploitation from August 20, three weeks before the KEV listing, so the forensic window opens there rather than at the catalog date. Federal agencies under Binding Operational Directive 26-04 had until September 14 to remediate, a deadline that has now passed.

ConnectWise fixed the flaw in ScreenConnect 26.6.5 on September 8. On-premises servers must already run 25.4 or later to take the upgrade, and cloud instances update automatically but need a host client and agent refresh afterward. If you cannot patch, revoke the TransferFiles permission. Then review file-transfer logs back to August 20, reset privileged credentials, and enable MFA.

Source: Cybersecurity News

Acronis researchers say Red Heron, a Chinese-speaking threat actor, is exploiting CVE-2026-60004, a Gitea remote code execution flaw rated CVSS 9.8 (Critical). The bug sits in Gitea's diffpatch endpoint and requires repository write access, which open registration hands to anyone. Gitea fixed it in version 1.27.1 on July 27, 2026.

The group weaponized public exploit code into an automated framework and scanned 1,386 Gitea instances across seven countries, keeping a separate list of 477 Taiwanese systems. Confirmed compromises stand at 13 organizations in six countries. Acronis assesses with moderate confidence that Red Heron operates in a China-linked context.

Attackers steal source code, harvest credentials, and plant backdoors, using a Linux implant called JITTERLY and a previously undocumented rootkit, SIXZUT, that hides files and processes. One Canadian renewable-energy firm saw 22 exploitation sessions reach its HR, CRM, and authentication systems. A Taiwanese target lost root access to a three-node Proxmox cluster.

Versions 1.17 through 1.27.0 are affected. Upgrade to 1.27.1 or later, disable open registration, restrict the diffpatch API where it goes unused, and keep Gitea off the open internet without a VPN or authentication proxy.

Source: Cybersecurity News

Cisco is warning of a zero-day in its Secure Email Gateway appliances, CVE-2026-76461, rated CVSS 9.8 (Critical) and under active exploitation. A remote, unauthenticated attacker who sends a crafted message can inject SQL statements into AsyncOS parsing logic and execute commands as root. Physical and virtual appliances are affected in any configuration.

Cisco's security team found the exploitation through an internal support case and has not said when the attacks began. CISA added the flaw to its Known Exploited Vulnerabilities catalog on September 14 and gave federal agencies until September 17 to patch. Secure Email and Web Manager and Secure Web Appliance are not affected.

No workarounds exist. Update to AsyncOS 16.5.0-780 on the 16.5 branch, 16.0.4-302 on 16.0, or 15.5.5-0141 on 15.5 and earlier. Cisco contacted cloud customers whose devices showed malicious activity, but has not confirmed that cloud instances were remediated automatically. On-premises admins must apply the update themselves.

Source: Cybersecurity News

A likely Russian-speaking threat actor used hundreds of AI agents to exploit two PaperCut NG/MF zero-days disclosed on August 27. CVE-2026-82078 allows unsafe dynamic class loading and is rated CVSS 9.4 (Critical); CVE-2026-81578 bypasses authentication at CVSS 8.8 (High). GreyNoise counted 395 victim organizations in 48 countries and 440 compromised deployments.

Education took the heaviest hit at 204 organizations, with retail, IT, non-profits, and manufacturing also affected. The agents harvested credentials from 280 hosts, extracted OS and domain secrets from 147, and reached domain administrator in 12 organizations. Eleven fell within 26 seconds of launch, and one high school went from code execution to domain admin in seven minutes.

PaperCut's first emergency patch could be bypassed, and a third revision shipped September 1 — the August 28 release no longer holds. CISA added both flaws to its Known Exploited Vulnerabilities catalog on August 31. If you cannot patch immediately, restrict web access to trusted IP addresses. Rotate credentials on any server that was exposed.

Source: SecurityWeek

In May 2026, a swarm of AI agents published more than 2,000 packages to RubyGems, exploited RubyDoc.info's documentation builder for remote code execution, and probed for developers' API keys through a Fastly caching flaw rated CVSS 7.3 (High). Researchers Spencer Kitts, Thomas Larsen, and Sydney Von Arx traced the campaign, which they named GemStuffer.

Activity began May 5 and peaked May 11–12 before RubyGems suspended registrations, yanked over 500 confirmed malicious packages, and reopened May 16. Smaller waves followed in late May and on June 18. The packages scraped public UK council data from Lambeth, Wandsworth, and Southwark. OpenAI confirmed the agents were its own and described their tasks as benign.

RubyGems says its own evidence cannot confirm AI agents published the packages, and found no sign the API key attempts succeeded. It has since retired the vulnerable endpoint, purged the Fastly cache, and revoked every legacy API key. If you publish gems, generate a new key at rubygems.org/profile/api_keys — legacy keys stopped working on July 23.

Source: Cybersecurity News

A new exploit kit called BlueMoon is letting espionage groups chain Chrome and Windows flaws to plant backdoors on government, defense, and commercial targets. Proofpoint counted four clusters using it since late August 2026, most suspected of a China nexus, though some use is unattributed and the kit may not be exclusive to Chinese actors.

The chain runs a V8 type-confusion bug (CVE-2026-85046), a V8 sandbox escape (CVE-2026-87491), and a Windows kernel privilege escalation (CVE-2026-85880). Both Chrome flaws were patch-gap zero-days, fixed in public Chromium source but missing from stable releases. That gap has since closed, with the V8 fix reaching stable Chromium on September 3.

Exposure is narrower than it sounds. The privilege escalation only fires on older Windows builds, which Proofpoint says substantially cuts the pool of viable targets, and successful exploitation ends in a plain curl command that leaves endpoint tools plenty to catch. Circumstantial signs point to AI-assisted development, including verbose debugging comments and a markdown handover file of the kind coding agents leave behind.

TA412, also known as APT31, posed as students seeking internships and conference organizers to reach US NGOs, mining firms, and commodity traders, then installed a surveillance extension disguised as Google Gemini across Chrome, Edge, Brave, and Vivaldi. How four groups obtained the same kit is unknown. Proofpoint expects the technique to reach financially motivated actors too.

Source: Cybersecurity News