Posts in Security （60）

* Image is for illustration purposes.

OpenAI Halts Frontier Model Training After AI Agent Found a DNS Loophole

* Image is for illustration purposes.

OpenAI to Unveil GPT-6 Cyber Security Model and New Defense Product at DevDay

* Image is for illustration purposes.

Google, Anthropic, and OpenAI Roll Out Cybersecurity-Focused AI Models in Quick Succession

* Image is for illustration purposes.

Cohere Adds Confidential Computing to Model Vault So It Can't See Customer AI Data

* Image is for illustration purposes.

Gyazo Breach Exposes 23.62 Million User Records and 490 Million Image Metadata Entries

* Image is for illustration purposes.

AI Agents Breach 395 Organizations Through PaperCut Flaws, Compromising 11 in 26 Seconds

* Image is for illustration purposes.

Attackers Are Draining Claude Usage Quotas by Stealing Login Sessions with Infostealer Malware

* Image is for illustration purposes.

Chrome Patches Its Sixth Zero-Day of 2026 with 152.0.7977.82, Closing a V8 Type Confusion Flaw

* Image is for illustration purposes.

CrowdStrike and OpenAI Expand Partnership: Codex Agents Watched at Runtime, GPT-5.6 Cyber Joins the Defense

Trezor Breach Now Affects About 81,000 People as 2019-2021 Orders Resurface

* Image is for illustration purposes.

Tenable Unveils Exchange Inspector to Vet AI Agents and MCP Servers Before They Ship

* Image is for illustration purposes.

Proofpoint Launches SOC Analyst Agent, Using OpenAI Daybreak Models to Investigate Threats While Containment Stays Human

* Image is for illustration purposes.

NEC Launches AI-Driven Managed Service That Handles Vulnerabilities From Discovery to Remediation

* Image is for illustration purposes.

OpenAI Launches Daybreak for Frontline Defenders, Committing $1 Billion to Cyber Defense for Water, Power, and Community Banks

* Image is for illustration purposes.

METR Loses About 93 Million Yen Worth of AI Credits to a Stolen API Key, Unnoticed for Three Weeks

* Image is for illustration purposes.

Google Donates Longfellow Zero-Knowledge Proof Library, Aiming at a Quantum-Safe Standard for Age Verification

* Image is for illustration purposes.

Roughly 5,000 Dropbox Accounts Hijacked as Lenovo ID Linking Becomes a Side Door

KAIST Builds SweepLED, a $7 LED Phone Case That Finds Hidden Cameras with 94% AI Accuracy

Meta Disables Cameras on Thousands of AI Glasses With Tampered LEDs, Detection Update Shuts Off Capture

* Image is for illustration purposes.

OpenAI Formally Designates Astra as Its First Critical-Level Cyber Model, Release Coming Soon with Advanced Features Gated Behind Daybreak Blue

Google Launches Fairwind Program for Cyber Defense, Bringing Gemini 3.8 Flash Cyber to Automated Vulnerability Patching

* Image is for illustration purposes.

AIR Raises 50 Million USD to Vet the Skills AI Agents Install, Blocking 27% of What It Finds

CrowdStrike Unveils SafeMind, an Agentic Cyber Defense System Built on NVIDIA Nemotron

* Image is for illustration purposes.

Anthropic Confirms Claude Session Hijacking as Infostealer Malware Drains Paid Usage