Malicious ChatGPT Share Links Could Expose Private Conversations
Researchers discovered a ChatGPT vulnerability that allowed malicious shared links to execute code in logged-in sessions and access private conversations. OpenAI fixed the flaw on October 26.
Critical UEFI PXE Flaw Can Run Code Before Secure Boot
A critical EDK II PXE flaw could let an unauthenticated attacker execute code during network boot, before Secure Boot checks the operating system.
U.S. Officials Suspect Iran-Linked Hackers in Military Freezer Attacks
U.S. officials reportedly suspect Iran-linked hackers behind August freezer failures at military commissaries, though investigators have not confirmed attribution or the intrusion method.
Pentagon Investigates Military Base Freezer Failures as Suspected Cyberattack
The Pentagon is investigating August freezer malfunctions at at least six U.S. military commissaries as a suspected cyberattack after cooling units spoiled food.
Discover-Tech Suspected of Tracking Starlink Users Through Ads
TrendAI found SHADOW-EARTH-79 unusually dominant in Starlink ad auctions across 105 countries. The bidder was later identified as Discover-Tech, raising concerns about surveillance through ad-tech data.
ESET Links UAC-0099 and Sandworm to Destructive Ukraine Cyberattack
ESET says UAC-0099 handed access to Sandworm in a 2025 attack on a Ukrainian grain company, where a new backdoor used the Yggdrasil mesh network for command-and-control.
OpenAI Reverses ChatGPT Pro Usage Cut, Doubles Work and Codex Limits
OpenAI canceled a planned Pro 200 Work and Codex allowance reduction and doubled included usage for Pro 100 and Pro 500 subscribers.
Microsoft Fixes Windows 11 Always On VPN Issues in October Preview
Microsoft's October 28 preview update KB5127936 fixes Always On VPN connection failures on Windows 11 25H2 and 26H2 after September security updates.
Tailscale 1.104.4 Fixes SSH and Serve Flaws That Could Expose Root Access
Tailscale 1.104.4 fixes four groups of SSH and Serve security flaws that could expose root-only sockets, enable node takeover or disrupt mesh connections.
New robots.txt Proposal Aims to Simplify AI Training Opt-Outs
An IETF draft proposes a shared robots.txt signal for AI training preferences. OpenAI says it will honor the signal, while Google plans support in 2027.
How to Stop WhatsApp From Automatically Downloading Photos and Videos
Stop WhatsApp from automatically downloading photos, videos, audio, and documents. Follow these steps on Android and iPhone to save data and storage.
Hudson Rock Details Malicious Script Behind Polymarket Attack
Hudson Rock found malicious code in a DebugBear-hosted script that tried to use active Polymarket sessions to move assets. The total losses remain unverified.
Anthropic Says Claude Built Exploits Across Five Platforms
Anthropic says an unreleased Claude model built five exploit chains across Chrome, Safari, Android, Windows and Linux, including a remote Linux root exploit.
Loading
