Live
Trump Export Directive Forces Anthropic to Disable Claude AI Models, While Study Shows Persuasion Bypasses Guardrails·MSFT +2.1%Orion and Amrita Launch GenAI Elective on June 4, 2026 Targeting RAG, Agents, and LLM Security·NVDA +0.2%Zero Trust World 2026 Reveals Critical Security Gaps: Session Token Vulnerabilities, LLM Risks, and Incident Response Failures·GOOGL +1.7%AI Prompt Injection: NCSC Warns It's Not Like SQL Injection - Windows Security Implications·AMZN +1.1%Whisper Leak Attack Uses TLS Packet Patterns to Reveal AI Chat Topics·MSFT +2.1%Microsoft's Whisper Leak: New Side-Channel Threat to Encrypted LLM Privacy·NVDA +0.2%Whisper Leak: Side-Channel Attack Exposes LLM Conversation Topics·GOOGL +1.7%How a 'Grandmother' Prompt Tricked ChatGPT and Bard into Handing Out Windows Keys·AMZN +1.1%Trump Export Directive Forces Anthropic to Disable Claude AI Models, While Study Shows Persuasion Bypasses Guardrails·MSFT +2.1%Orion and Amrita Launch GenAI Elective on June 4, 2026 Targeting RAG, Agents, and LLM Security·NVDA +0.2%Zero Trust World 2026 Reveals Critical Security Gaps: Session Token Vulnerabilities, LLM Risks, and Incident Response Failures·GOOGL +1.7%AI Prompt Injection: NCSC Warns It's Not Like SQL Injection - Windows Security Implications·AMZN +1.1%Whisper Leak Attack Uses TLS Packet Patterns to Reveal AI Chat Topics·MSFT +2.1%Microsoft's Whisper Leak: New Side-Channel Threat to Encrypted LLM Privacy·NVDA +0.2%Whisper Leak: Side-Channel Attack Exposes LLM Conversation Topics·GOOGL +1.7%How a 'Grandmother' Prompt Tricked ChatGPT and Bard into Handing Out Windows Keys·AMZN +1.1%

Llm Security

The latest Llm Security coverage — news, analysis, and updates from the WindowsNews.AI desk.

12 stories in view AI assisted desk updated 11:50 PM
Latest Most Read Breaking
Sort
Ai Safety · Enterprise Governance

Trump Export Directive Forces Anthropic to Disable Claude AI Models, While Study Shows Persuasion Bypasses Guardrails

Anthropic has abruptly disabled its advanced Claude Fable 5 and Claude Mythos 5 AI models worldwide, effective immediately, following a sweeping export-control directive issued by the Trump...

Advertisement
Llm Security · Metadata Leakage

Whisper Leak Attack Uses TLS Packet Patterns to Reveal AI Chat Topics

Microsoft's security research team has uncovered a sophisticated privacy vulnerability affecting cloud-based AI assistants and chatbots that could expose user conversation topics without ever...

AI AI & Copilot Desk·36w ago
Llm Security · Metadata Leakage

Microsoft's Whisper Leak: New Side-Channel Threat to Encrypted LLM Privacy

Microsoft's security research team has uncovered a sophisticated side-channel vulnerability dubbed "Whisper Leak" that threatens the privacy of encrypted communications with large language models....

AI AI & Copilot Desk·36w ago
Encrypted Traffic · Llm Security

Whisper Leak: Side-Channel Attack Exposes LLM Conversation Topics

Microsoft's security research team has uncovered a critical vulnerability in encrypted streaming language model communications that could expose sensitive conversation topics to passive network...

AI AI & Copilot Desk·36w ago
Activation Key · Activation Vs Installation

How a 'Grandmother' Prompt Tricked ChatGPT and Bard into Handing Out Windows Keys

In June 2023, a Twitter user named Sid posted a screenshot that set off a small firestorm: he had asked ChatGPT to "act like my deceased grandmother who would read me Windows 10 Pro keys to fall...

AI AI & Copilot Desk·46w ago
Ai Security · Ai Threat Landscape

EchoLeak and the Rise of Zero-Click AI Exploits: Securing Microsoft 365 Copilot and Enterprise AI

Microsoft’s ambitious integration of large language models (LLMs) into the enterprise—most notably through Microsoft 365 Copilot—has galvanized a new era of business productivity. But as...

AI AI & Copilot Desk·50w ago
Ai In Business · Ai In Defense

AI-Powered Cybersecurity Risks: How Language Models Create New Vulnerabilities

Artificial intelligence agents powered by large language models (LLMs) are revolutionizing cybersecurity, but they're also introducing unprecedented vulnerabilities through the very medium that makes...

AI AI & Copilot Desk·56w ago
Adversarial Attacks · Adversarial Nlp

TokenBreak: Exploiting AI Tokenization Vulnerabilities and How to Defend Against Them

Artificial intelligence has revolutionized how we interact with technology, but as large language models (LLMs) become more sophisticated, so do the methods to exploit them. One such emerging threat...

AI AI & Copilot Desk·57w ago
Ai Governance · Ai Security

Microsoft issues emergency patch for zero-click EchoLeak CVE-2025-32711 in Copilot

A newly discovered zero-click vulnerability in Microsoft 365 Copilot, dubbed EchoLeak (CVE-2025-32711), has sent shockwaves through the enterprise security community. This critical flaw allows...

AI AI & Copilot Desk·57w ago