LIVE · cybersecurity feed
Live wire
CVE-2026-107441 · Zyxel GS1900 Series Switches Vulnerability Exploited On Day Of DisclosureCVE-2026-62032 · Designinvento DirectoryPress Flaw Added To VulnCheck KEV After 28 DaysCVE-2024-1775 · Nextend Social Login Vulnerability Added To VulnCheck KEV After 981 DaysCVE-2013-2165 · Red Hat JBoss RichFaces Vulnerability Added To VulnCheck KEV After 13 YearsCVE-2026-15760 · Divi Essential Plugin Missing Authorization Flaw ExploitedOpenRouter Drops 3 Providers After Researchers Catch Months of Prompt Theft at AI Hosting FirmsCVE-2026-125396 · Researchers Demonstrate Ghostscript Sandbox-Escape RCE via Crafted File Uploads in Popular AppsCVE-2024-3848 · LFProjects MLflow Path Traversal Vulnerability Added To VulnCheck KEVCVE-2026-118267 · WPDataAccess Incorrect Authorization Flaw Exploited Same Day As DisclosureCVE-2026-118066 · LatePoint Improper Authentication Vulnerability Exploited On Day Of Disclosure
nation-state

Signal to block Siri, Gemini AI from capturing chats on iOS, Android

Signal has implemented new security measures designed to prevent AI assistants, specifically Apple’s Siri and Google’s Gemini, from accessing or capturing private chat content displayed within the application. These protections are being deployed across iOS and Android platforms to address concerns that OS-level AI features, such as Siri’s Onscreen Awareness and Google’s Gemini Intelligence,…

ZeroDay News ·

Source: BleepingComputer

Signal has implemented new security measures designed to prevent AI assistants, specifically Apple’s Siri and Google’s Gemini, from accessing or capturing private chat content displayed within the application. These protections are being deployed across iOS and Android platforms to address concerns that OS-level AI features, such as Siri’s Onscreen Awareness and Google’s Gemini Intelligence, could inadvertently expose sensitive user data by reading the contents of the screen.

The organization confirmed that these updates follow a similar initiative launched in May 2025, which introduced protections against the screenshot-capturing capabilities of Microsoft’s Recall feature on Windows 11. Signal’s engineering team spent several months analyzing how these AI systems interact with application data and reverse-engineering the extraction pathways to develop targeted blocks. While the new protections are designed to trigger automatically when the application detects an AI assistant attempting to scan the screen, the company has not provided specific version numbers for the update, advising users to ensure their application is fully updated to the latest release.

On Android devices, the application will automatically activate its Screen Security feature upon detecting an AI assistant attempt to access the display. For iOS users, the implementation is more restrictive; enabling the protection against Siri AI also prevents the user from taking manual screenshots of their own conversations. Signal recommends that iOS users navigate to the App Security section within the application settings to verify that the block is active.

Despite these efforts, Signal acknowledges that the integration of AI agents deep within modern operating systems presents an ongoing challenge. The company warns that these agents may still find alternative methods to access conversation data, which it describes as an existential threat to the privacy model of the platform. Signal explicitly noted that it currently lacks the capability to provide similar screenshot-blocking protections on macOS 27. For users on that platform, the company suggests manually adding Signal to the Excluded Apps list within the system settings for Siri to prevent the assistant from proactively accessing message content.

The broader concern raised by the organization is the lack of transparency from major operating system developers. Signal is calling on Apple, Google, and Microsoft to provide comprehensive documentation regarding how AI agents interact with application data. Furthermore, the company is urging these tech giants to offer developers clear, standardized mechanisms to opt out of AI access across various system components, including screens, file systems, notifications, and application programming interfaces.

Signal maintains that while it has successfully blocked current known extraction pathways, the evolving nature of agentic systems means that application-level defenses have inherent limitations. By proactively restricting access, the company aims to keep private data within the control of the user, yet it emphasizes that true security in the age of integrated AI requires greater cooperation and disclosure from the providers of the underlying operating systems.

nation-stateai
ShareXLinkedInWhatsAppFacebook

More News

view all →
vulnerability

SECURITY AFFAIRS AI-CYBERSECURITY NEWSLETTER ROUND 9

The cybersecurity landscape has faced a surge in automated threats and critical vulnerability disclosures throughout late November 2026. A primary driver of this activity is the integration of artificial intelligence into offensive operations, with researchers observing the use of the Cairn AI framework to facilitate autonomous zero-day exploitation. Furthermore, threat intelligence analysts…

malware

SECURITY AFFAIRS MALWARE NEWSLETTER ROUND 123

The cybersecurity landscape has faced a surge of activity in late November 2026, characterized by a mix of active exploitation of known vulnerabilities, sophisticated supply chain compromises, and emerging threats involving artificial intelligence. The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has significantly expanded its Known Exploited Vulnerabilities (KEV) catalog,…

zero-day

Security Affairs newsletter Round 605 by Pierluigi Paganini – INTERNATIONAL EDITION

The cybersecurity landscape has faced a surge of activity involving both sophisticated AI-driven exploitation and widespread supply chain compromises. Recent reports confirm that the Norwegian energy giant Equinor is currently managing a cyberattack against its IT infrastructure, though the company has stated that oil and gas production remains unaffected. Simultaneously, the European…

CVE-2026-107441high

Zyxel GS1900 Series Switches Vulnerability Exploited On Day Of Disclosure

The Zyxel GS1900 Series Switches Out-of-bounds Write vulnerability was identified as exploited on the same day it was published, yet it remains absent from major government catalogues.

CVE-2026-62032high

Designinvento DirectoryPress Flaw Added To VulnCheck KEV After 28 Days

The DirectoryPress plugin for WordPress contains a local file inclusion vulnerability that was added to the VulnCheck KEV catalogue 28 days after its initial publication.

CVE-2024-1775high

Nextend Social Login Vulnerability Added To VulnCheck KEV After 981 Days

The Nextend Social Login plugin vulnerability CVE-2024-1775 has been added to the VulnCheck KEV catalogue, nearly three years after its initial publication.