Today’s briefing covers 15 security issues reported in the last 24 hours, grouped by type: 2 actively exploited, 4 breaches and data exposure, 5 ransomware and attacks, 4 patches and advisories. Each entry says who is affected and what to do, with links to the original reports.<\/p>\n
Who is affected:<\/strong> All on-premises VeloCloud Orchestrator releases before 5.2.3.18, 6.1.3.10 and 6.4.2.12 were affected, and no configuration can prevent exposure.<\/p>\n Attackers exploited a remotely exploitable SQL injection, tracked as CVE-2026-114625 with a CVSS score of 9.9, to obtain operator privileges and access privileged functionality. Arista has patched hosted deployments, while attacks against exposed on-premises systems originated from two identified IP addresses.<\/p>\n What to do:<\/strong> Update on-premises VCO installations as soon as possible, block the identified IP addresses, and preserve relevant logs and file-system timestamps before remediation where feasible.<\/p>\n Sources:<\/strong> securityweek.com<\/a>, CISA KEV \/ NVD<\/a><\/p>\n Who is affected:<\/strong> Organisations running Collabora Collabora Online, Built-in CODE Server. CISA added it to its Known Exploited Vulnerabilities list on 2026-10-23, which means attacks are confirmed.<\/p>\n Collabora Collabora Online, Built-in CODE Server when used within the NextCloud platform contains an OS command injection vulnerability that allows for remote code execution when the CODE server isn’t running.<\/p>\n What to do:<\/strong> Apply mitigations in accordance with vendor instructions, ensuring compliance with CISA\u2019s BOD 26-04 Prioritizing Security Updates Based on Risk (see URL in Notes) guidance and CISA\u2019s \u201cForensics Triage Requirements\u201d (see URL in Notes). Follow applicable BOD 26-04 guidance for cloud services or discontinue use of the product if mitigations are unavailable. Stakeholders are responsible for evaluating each asset’s internet exposure and ensuring adherence to BOD 26-04 patching guidelines. US federal agencies must fix it by 2026-10-26.<\/p>\n Sources:<\/strong> CISA KEV \/ NVD<\/a><\/p>\n Who is affected:<\/strong> All Railway users may be affected; the platform has around 4.8 million users and hosts more than 16 million services.<\/p>\n Railway detected suspicious activity involving systems that store customer environment variables and said an unauthorized third party may have obtained decrypted variables for some users. Exposed variables may contain API keys, database credentials, registry credentials, and other secrets.<\/p>\n What to do:<\/strong> Immediately rotate environment variables and stored secrets, including private registry credentials, and check associated service logs for unusual activity.<\/p>\n Sources:<\/strong> bleepingcomputer.com<\/a><\/p>\n Who is affected:<\/strong> State Bank of India customers represented by 3,897 leaked records may be affected; the source does not say whether each record belongs to a different customer.<\/p>\n An anonymous hacker posted a small batch of State Bank of India customer records containing personal information, new account and loan applications, and other data. The bank acknowledged the breach in a notice to India’s National Stock Exchange and described the impact as affecting a very small number of customers.<\/p>\n What to do:<\/strong> Affected customers should monitor their accounts and identity records for misuse and follow any further instructions from State Bank of India.<\/p>\n Sources:<\/strong> darkreading.com<\/a><\/p>\n Who is affected:<\/strong> Certain Quest Diagnostics patients may have had records accessed through the Quanum Lab Services Manager; the number of affected patients and the specific records remain unknown.<\/p>\n Quest Diagnostics reported unauthorized access to a platform that certain clients use to place test orders and receive or retrieve results. The company discovered the breach on Wednesday and notified federal law enforcement while investigating the scope and working with cybersecurity experts. Quest said the breach was unrelated to the PHP issue.<\/p>\n What to do:<\/strong> Watch for notices from Quest Diagnostics about affected records and follow the company’s identity protection or account security guidance if provided.<\/p>\n Sources:<\/strong> securityweek.com<\/a><\/p>\n Who is affected:<\/strong> The breach affected 11,774,247 CAPCOM ID users worldwide, including 196 users whose passwords were changed.<\/p>\n Attackers stole email addresses and account data from CAPCOM ID, including creation dates, login details and last-login IP addresses. For 43 users, dates of birth, gender, country or region, and nicknames were also compromised. Capcom found no evidence that customer passwords were stolen and has reset the 196 affected passwords.<\/p>\n What to do:<\/strong> Avoid URLs and attachments in unsolicited emails or messages that impersonate Capcom or related entities, and contact Capcom through its dedicated support channels if account activity is suspicious.<\/p>\n Sources:<\/strong> securityweek.com<\/a>, helpnetsecurity.com<\/a><\/p>\n Who is affected:<\/strong> Jared Kushner and the United Arab Emirates-based Watani Al Emarat Foundation are reported to have been affected through monitoring of Kushner’s iPhone X and access to an Alpha bank account.<\/p>\n Dutch intelligence reportedly captured and turned a Russian GRU hacker in late 2017, gaining access to a server containing iPhone screenshots and the method used to penetrate the device. Intelligence personnel then reportedly monitored Kushner’s phone from early 2018 and saw him access a Dubai-based account that had received around $750 million.<\/p>\n What to do:<\/strong> Users responsible for sensitive communications should use approved secure devices and channels and review how personal phones, accounts, and credentials are protected.<\/p>\n Sources:<\/strong> gizmodo.com<\/a><\/p>\n Who is affected:<\/strong> The Prince of Persia APT targets Iranian dissidents and government and private entities across the US, Europe, and the Middle East.<\/p>\n The Iran government-sponsored group resumed operations after three months with a command-and-control server on US shared hosting, its first such server since 2016. Its Tonnerre malware stole more than 1,800 files from one victim within a week, while its Tornado malware uses a Telegram bot to steal files.<\/p>\n What to do:<\/strong> Defenders should monitor suspicious Telegram traffic and check systems against the published indicators of compromise.<\/p>\n Sources:<\/strong> securityweek.com<\/a><\/p>\n Who is affected:<\/strong> PHP servers exposed to the internet are affected, with exploit attempts reaching the millions; no successful exploitation was observed in the wild.<\/p>\n Attackers attempted to exploit CVE-2026-109618 and CVE-2026-109619 to deploy a Mirai variant called Potassium, but the attempts were unsuccessful. The vulnerabilities enable unauthenticated remote code execution and were patched in PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35.<\/p>\n What to do:<\/strong> Update PHP to a patched version and check servers for indicators associated with the Potassium campaign.<\/p>\n Sources:<\/strong> thehackernews.com<\/a>, thenextweb.com<\/a><\/p>\n Who is affected:<\/strong> Users who downloaded malicious LiteLLM 1.107.0 packages during the compromise window may have installed an information stealer and remote access trojan; the packages were downloaded over 49,000 times, while enterprise customers using the private repository were not affected.<\/p>\n Attackers used compromised GitHub CLI OAuth tokens to change the release workflow, publish artifacts from a lookalike repository and serve poisoned packages on PyPI and Docker. The malicious packages were online for roughly three hours and removed or replaced with clean artifacts. LiteLLM’s official GitHub source tag for version 1.107.0 did not contain the malicious files.<\/p>\n What to do:<\/strong> Remove LiteLLM 1.107.0 downloaded from PyPI or Docker during the exposure window, rotate exposed credentials and check systems for the stealer and remote access trojan.<\/p>\n Sources:<\/strong> securityweek.com<\/a>, thenextweb.com<\/a><\/p>\n Who is affected:<\/strong> Owners of internet-accessible self-hosted models, AI gateways, internal test systems and Model Context Protocol servers may have their models and computing resources abused.<\/p>\n An AI attack tool scans for reachable models and directs them to find new targets, exploit common web frameworks, steal databases and credentials and install cryptocurrency miners without human approval. A honeypot recorded more than 127,000 requests from tens of thousands of unique addresses, mostly reconnaissance, stolen API key attempts and exposed configuration probes.<\/p>\n What to do:<\/strong> Restrict model and MCP endpoints from public access, rotate exposed API keys and review accounts and systems for unauthorized activity.<\/p>\n Sources:<\/strong> helpnetsecurity.com<\/a><\/p>\n Who is affected:<\/strong> Switch 2 users and developers running homebrew, reverse engineering, mods, save backups, or peripheral access may be affected.<\/p>\n A software-only, tethered partial jailbreak gained userland access to the Switch 2 through an integer overflow in the original Switch version of SmileBasic 4. It also required a macOS or Linux computer on the same network to run a Node.js package that enables privileged system access.<\/p>\n What to do:<\/strong> Do not run the demonstrated exploit or related tooling on Switch 2 systems until the security flaw and appropriate mitigations are addressed.<\/p>\n Sources:<\/strong> arstechnica.com<\/a><\/p>\n Who is affected:<\/strong> Atlassian Bamboo, Bitbucket, Confluence, Crowd, Fisheye\/Crucible, Jira Software and Jira Service Management deployments are affected; the bulletin covers 75 unique CVEs, six rated critical.<\/p>\n Atlassian’s October 2026 bulletin lists 176 vulnerability entries tied to third-party dependencies, including 162 high-severity and 14 critical-severity issues. No exploitation was reported. Atlassian says the affected component use presents lower, non-critical assessed risk for its customers.<\/p>\n What to do:<\/strong> Update Atlassian products to the latest versions or the fixed releases listed in the October bulletin.<\/p>\n Sources:<\/strong> securityweek.com<\/a><\/p>\n Who is affected:<\/strong> The risk affected agents using x402 SDKs that automatically signed payments when their applications set no spending limits.<\/p>\n The x402 SDK default client initially had no spending limit and selected the first payment option offered by a server, allowing a malicious API to request an agent’s full supported token balance. Releases published on August 18 added a limit, and the x402 Foundation had fixed all three critical flaws found in the code. No attacks using the flaws had been publicly reported.<\/p>\n What to do:<\/strong> Check whether the TypeScript x402 SDK is in use and upgrade to a release containing the August 18 fixes, then confirm that spending limits are configured.<\/p>\n Sources:<\/strong> thehackernews.com<\/a><\/p>\n Who is affected:<\/strong> The flaw affects applications on Node.js 22.x, 24.x and 26.x that decode untrusted Brotli data, including through built-in fetch responses or middleware decoding request bodies.<\/p>\n Researchers demonstrated that crafted Brotli data could execute attacker-supplied code on an official Node.js 24.22.0 Linux arm64 build through the bundled Brotli decoder. Node.js released fixes on October 21, and version 24.22.1 rejected the test data without running code. No exploitation in the wild had been reported as of October 22.<\/p>\n What to do:<\/strong> Update applications that decode untrusted Brotli data to a fixed Node.js release.<\/p>\n Sources:<\/strong> thehackernews.com<\/a><\/p>\nCollabora Collabora Online, Built-in CODE Server: Collabora Online, Built-in CODE Server OS Command Injection Vulnerability (CVE-2026-110447)<\/h3>\n
Breaches and data exposure<\/h2>\n
Railway warns users to rotate secrets after suspected exposure<\/h3>\n
State Bank of India discloses leak of 3,897 customer records<\/h3>\n
Quest Diagnostics Breach Exposes Certain Patient Records<\/h3>\n
Capcom breach exposes CAPCOM ID data for 11.7 million people<\/h3>\n
Ransomware and attacks<\/h2>\n
Dutch intelligence reportedly accessed Kushner’s iPhone through GRU hack<\/h3>\n
Iran’s Prince of Persia APT returns with US-hosted command server<\/h3>\n
PHP2Shell Exploit Attempts Target Unpatched PHP Servers<\/h3>\n
Poisoned LiteLLM 1.107.0 Packages Served on PyPI and Docker<\/h3>\n
LLMjacking Tool Hijacks Exposed AI Models for Attacks<\/h3>\n
Patches and advisories<\/h2>\n
Switch 2 partial jailbreak exploits SmileBasic integer overflow<\/h3>\n
Atlassian Patches Third-Party Flaws Across Server Products<\/h3>\n
x402 SDK flaw let malicious APIs request an agent’s full token balance<\/h3>\n
Node2Shell flaw allowed code execution through crafted Brotli data<\/h3>\n
Sources<\/h2>\n