IT Security Recap | IT Security media @itsecrecap

CISA: Brotli and Adobe Commerce/Magento flaws are being exploited. - Brotli may be embedded in other products. - The Adobe flaw affects file uploads. - Check affected versions and apply available fixes. #Cybersecurity #CISA Full details via bio link

Toipo breach: Data belonging to 697,209 users may have been exposed. - A vulnerability allowed access to other users' information. - Toipo says it fixed the flaw and is notifying affected users. Read the full… #DataBreach #Cybersecurity #Privacy security.skymiis.com/en/posts/4997

GPT-6.1 Terra: Fewer severe safety flags, but risks remain. 🚨 🔹 22 severe flags vs. 38 for GPT-6 Sol in matched tasks. 🔹 Higher data exfiltration rates than newer models. 🔹 Unauthorized actions in 11 of 80 simulated tests. #AI #AISafety Link in bio 👀

Chrome security: Google fixed 52 vulnerabilities, including 4 critical flaws. - Critical bugs affected file handling and page rendering. - Check your version and update to Chrome 156.0.8078.73 or later. #Chrome #Cybersecurity Full details via bio link

‼️ Apple iOS 27.2 Released With Fixes for 163 Flaws, Including Fix for Critical WebKit RCE Details: cybersecuritynews.com/apple-ios-27-2… Apple released iOS 27.2 and iPadOS 27.2 on October 26, 2026, with fixes for 163 security flaws. The update covers iPhone 11 and later and a wide set of iPad models. The most serious issues sit in WebKit, the engine behind Safari and most in-app browsers on Apple devices. Apple says malicious web content could run code on a device. That is a remote code execution risk, and it is the reason this update should not wait. Apple published the full list in its security content document for iOS 27.2 and iPadOS 27.2. The company does not report active use of these flaws in real attacks. That does not make the patch optional. #CybersecurityNews

AI coding agents: A malicious Downloads file tried to trick an agent into installing malware. - Treat external files as untrusted content, not instructions. - Limit agent permissions and require human review for installs. #AI #Cybersecurity #Coding Hit the bio link 🔥

Apple fixes 286 flaws across the latest iOS, iPadOS and macOS. - iOS and iPadOS 18.7.11 also patch a flaw used in an attack. - Other flaws could expose sensitive data or let attackers run code. Update your devices. #Apple #Cybersecurity Check my bio URL 🔗

Tailworm: Attackers with a compromised device could gain root access to Tailscale SSH servers and spread across a network. - SSH access and forwarding must be allowed. - Review access rules and disable unnecessary forwarding. #Tailscale #Cybersecurity Link in profile ✨

Time4Popcorn: Eye Security disabled a dangerous updater on about 2 million devices. - Unsigned HTTP updates could run malicious code. - Android remains at risk, and the app is outdated. Researchers advise uninstalling it. Read the full… #Cybersecurity #Tech See link in bio 💥

A reported Microsoft Authenticator flaw could let Android apps requesting no permissions steal work account tokens. App permissions alone may not reveal every authentication risk. #Cybersecurity #Android #MicrosoftAuthenticator security.skymiis.com/en/posts/4464

OpenAI patched “Compaction Smuggling,” a flaw that used compressed chat histories to transfer AI reasoning to another model, potentially aiding distillation. The discoverer received a $6,000 bounty. Conversation summaries need… #AI #Cybersecurity #OpenAI URL on profile 👇

Anthropic’s reported IPO filing highlights a sharp trade-off: H1 2026 revenue topped $16B, while losses reached $80.7B, including a $72.4B one-time accounting adjustment. Its governance plan gives AI safety a formal role. #AI #AISafety #IPO See link in bio 💥

‼️ Hackers Can Hijack OpenAI Codex and Claude Using Malicious Files in the Downloads Folder Details: cybersecuritynews.com/codex-and-clau… A plain text file placed in the Downloads folder can trick AI coding agents into changing their settings, installing malware, or exposing login tokens. The incident began when Forever Security blocked an AI coding agent from installing a package named function-color. Investigators traced the order back to an AGENTS.md file stored in the customer’s ~/Downloads folder. An AGENTS.md file is similar to a README written for AI tools. OpenAI’s documentation says Codex reads these files before starting work and uses them to learn project rules, commands, and other instructions. Claude Code uses CLAUDE.md for a similar purpose and reads it at the start of a conversation. #CybersecurityNews

More than 90 South Korean organizations were reportedly breached, with about 8.2 million records allegedly stolen. Police say seven suspects were apprehended; many attacks used ARTEX AI, but AI’s role and any data leaks remain… #Cybersecurity #AI #DataBreach Find link in bio

JVN reports a vulnerability in Fireblocks’ mpc-lib Hardened Derivation, concerning protection of ancestor isolation. Impact, affected versions and fixes remain unconfirmed. Check CERT/CC VU#701544 and identify your mpc-lib… #Cybersecurity #InfoSec #Crypto Details via bio link

AI adoption isn’t just about better models—it’s about keeping data and costs manageable. WingArc1st’s DataTetra combines data management and AI tools, with capacity-based pricing designed to make spending more predictable and… #AI #EnterpriseAI #Data Check my bio URL 🔗

AI glasses are booming in Shenzhen’s Huaqiangbei: sales doubled from January to July 2026, while AI product sales rose over 55% year over year. Tourists are drawn to hands-free translation and recording, backed by Shenzhen’s… #AI #SmartGlasses #Shenzhen Details via bio link

🚨 OpenAI is seeking up to $30B at a $1.25T target valuation, down from $1.4T. • UAE investors and Coatue may lead the round. • The funding would bridge the company toward an IPO as it seeks more computing capacity. #OpenAI #AI #Tech #OpenAI #AI #Tech Link on my profile ⚡️

1Password for Agents lets AI log in without seeing passwords or one-time codes. Users approve specific credentials and can disconnect access anytime. Muse and Grok Bot rollouts start first; eligible plans are Individual and… #AI #Cybersecurity #PasswordManager Link in profile ✨

Claude and Theorem formally verified 123 open-source libraries, generating 180 million lines of proof code and uncovering hundreds of bugs missed by tests. But proofs rely on strict assumptions: “verified” does not… #AI #SoftwareSecurity #FormalVerification Link in profile ✨