Daily brief: Friday 23 October 2026
Attackers were trying the PHP2Shell flaws days before the patches, LiteLLM's release pipeline was hacked to ship a credential stealer, GitLab has fixed another maximum-severity flaw, and Microsoft says its Always On VPN fix didn't work after all.
PHP2Shell flaws were being exploited four days before the fixes
Amazon Threat Intelligence says it saw an exploitation attempt against the two critical PHP flaws, CVE-2026-109618 and CVE-2026-109619, on 17 October, four days before public disclosure. On 21 October it saw several proof-of-concept exploits from about 02:00 UTC, roughly 12 hours before the 14:00 UTC patch release. PHP's fixed versions are 8.5.12, 8.4.27, 8.3.36 and 8.2.35. AWS says there is no configuration-only workaround that fully covers both flaws.
Source: Security Affairs
LiteLLM release pipeline hacked to ship a credential stealer
Attackers compromised LiteLLM's release pipeline on 21 October and published a malicious version 1.107.0 to PyPI, Docker Hub and GHCR for a few hours. The malware adds startup files to the Python library folder, so it runs whenever Python or pip starts. It goes after cloud and deployment credentials, AI provider keys, private keys, browser data and shell histories. The project says its private enterprise version was not affected.
Source: BleepingComputer
GitLab fixes another maximum-severity code execution flaw
GitLab has fixed CVE-2026-110264, an integer overflow in its Regular Expression Compiler that could let an unauthenticated user run code on the server. Fixed versions of Community and Enterprise Editions are 19.5.1, 19.4.3, 19.3.6, 19.2.9, 19.1.9, 19.0.10 and 18.11.13. There are no reports of it being exploited. GitLab.com is already patched.
Source: BleepingComputer
Microsoft says October updates didn't fix the Always On VPN bug
September's Windows security updates broke some Always On VPN connections on Windows 11 24H2 and later. Microsoft marked the problem resolved on 13 October, then changed it back to mitigated because it wasn't fixed. Affected connections use automatic protocol selection or fallback between IKEv2 and SSTP, and can get stuck on 'Connecting' or show 'The specified port is already in use'. A permanent fix is due 'in the coming weeks', with no date.
Source: BleepingComputer
Railway tells users to rotate secrets after a possible breach
Cloud hosting platform Railway found suspicious activity on systems that store customer environment variables. It says a third party may have accessed encrypted environment variables and, for some, the decrypted contents, though only a subset of users seems affected. Railway says it was part of a series of attacks on AI inference providers, and it paused deployments while it investigated.
Source: BleepingComputer
Rapid7: serious flaws are exploited within three days on average
Rapid7's Q3 2026 report says high-severity vulnerabilities were exploited an average of three days after disclosure, and 71% within seven days. It says 79% of them needed no credentials and no user action. The most active ransomware groups by claimed victims were The Gentlemen and Qilin. Rapid7 says AI played an operational part in 8% of incidents, with humans still choosing the targets.
Source: SecurityWeek
Anthropic launches Claude Duet, a proactive AI agent
Anthropic has launched Claude Duet, an agent that connects to your apps, sends a short morning brief and checks in every few hours with alerts. It runs in the cloud with its own browser and asks for approval before sending messages, making payments, signing documents or deleting files. It is a research preview on the web and iOS, by waitlist, for Pro and Max subscribers in the United States only.
Source: MacRumors
Weather
Live forecastLoading the forecast…