Ai-Security

OpenAI Codex had two sandbox escapes — both now patched

OpenAI Codex had two sandbox escapes — both now patched

Sep 20, 2026 · 2 min read

BragJack: Malicious Extensions Can Hijack AI Browser Agents

BragJack: Malicious Extensions Can Hijack AI Browser Agents

Sep 19, 2026 · 1 min read

Researchers Used Claude Opus 5 to Compromise OpenAI Accounts

Researchers Used Claude Opus 5 to Compromise OpenAI Accounts

Sep 19, 2026 · 0 min read

Google Gemini Accessed Real Company Systems During Security Test

Google Gemini Accessed Real Company Systems During Security Test

Sep 19, 2026 · 1 min read

Spain's Data Agency Receives First AI Agent Breach Report

Spain's Data Agency Receives First AI Agent Breach Report

Sep 17, 2026 · 1 min read

Seven Chinese AI Labs Caught in Industrial-Scale Claude Theft

Seven Chinese AI Labs Caught in Industrial-Scale Claude Theft

Sep 12, 2026 · 1 min read

Stolen AI Session Tokens Can Bypass MFA, Okta Research Finds

Stolen AI Session Tokens Can Bypass MFA, Okta Research Finds

Sep 10, 2026 · 2 min read

ChatGPT Flaw Let a Hidden Prompt Read and Send Your Gmail Data

ChatGPT Flaw Let a Hidden Prompt Read and Send Your Gmail Data

Sep 8, 2026 · 1 min read

GPT-6 Astra Aces Exploit Benchmarks; OpenAI Restricts It

GPT-6 Astra Aces Exploit Benchmarks; OpenAI Restricts It

Sep 4, 2026 · 1 min read

Malicious Git Configs Can Make AI Coding Agents Run Attacker Code

Malicious Git Configs Can Make AI Coding Agents Run Attacker Code

Sep 2, 2026 · 1 min read

Langflow Flaw CVE-2026-0768 Exploited to Steal AI API Keys

Langflow Flaw CVE-2026-0768 Exploited to Steal AI API Keys

Sep 1, 2026 · 1 min read

AI Safety Org METR Hit by API Key Theft Worth $600K in Credits

AI Safety Org METR Hit by API Key Theft Worth $600K in Credits

Sep 1, 2026 · 1 min read

CISA Flags Actively Exploited Ray AI Framework RCE Flaw

CISA Flags Actively Exploited Ray AI Framework RCE Flaw

Aug 19, 2026 · 1 min read

One-Click Copilot Flaw Let Attackers Silently Steal Data

One-Click Copilot Flaw Let Attackers Silently Steal Data

Aug 19, 2026 · 1 min read

Malicious MCP Servers Can Trick AI Coding Agents Into Leaking Your Secrets

Malicious MCP Servers Can Trick AI Coding Agents Into Leaking Your Secrets

Aug 12, 2026 · 1 min read

Atlassian Rovo AI Can Be Tricked Into Leaking Jira and Confluence Data

Atlassian Rovo AI Can Be Tricked Into Leaking Jira and Confluence Data

Aug 8, 2026 · 1 min read

Claude Code and Gemini CLI Flaws Leak CI Workflow Secrets

Claude Code and Gemini CLI Flaws Leak CI Workflow Secrets

Aug 7, 2026 · 1 min read

Hugging Face Diffusers Flaws Let Model Repos Execute Arbitrary Code

Hugging Face Diffusers Flaws Let Model Repos Execute Arbitrary Code

Aug 3, 2026 · 1 min read

Claude AI Escaped Its Sandbox and Uploaded Malware to PyPI

Claude AI Escaped Its Sandbox and Uploaded Malware to PyPI

Jul 31, 2026 · 1 min read

OpenAI Models Exploited Artifactory Zero-Days, Breached Hugging Face

OpenAI Models Exploited Artifactory Zero-Days, Breached Hugging Face

Jul 28, 2026 · 1 min read

Claude Cowork Flaw Let AI Agents Escape to Your Mac's Files

Claude Cowork Flaw Let AI Agents Escape to Your Mac's Files

Jul 23, 2026 · 1 min read

AI Agent Hacked Hugging Face, Stole Internal Credentials

AI Agent Hacked Hugging Face, Stole Internal Credentials

Jul 20, 2026 · 1 min read

Attackers Can Use Notifications to Hijack Google Gemini

Attackers Can Use Notifications to Hijack Google Gemini

Jun 3, 2026 · 1 min read

Anthropic's AI Finds 10,000 Critical Security Bugs Before Hackers Can

Anthropic's AI Finds 10,000 Critical Security Bugs Before Hackers Can

May 23, 2026 · 0 min read

Anthropic's MCP Protocol Design Flaw Potentially Exposes 200,000 Servers

Anthropic's MCP Protocol Design Flaw Potentially Exposes 200,000 Servers

Apr 18, 2026 · 0 min read

The Claude Code Source Leak: What Happened, What Was Exposed, and What It Means for You

The Claude Code Source Leak: What Happened, What Was Exposed, and What It Means for You

Apr 7, 2026 · 12 min read

CISA adds Langflow CVE-2026-33017 to KEV after active RCE exploitation

CISA adds Langflow CVE-2026-33017 to KEV after active RCE exploitation

Mar 27, 2026 · 2 min read

AI sandbox flaws in Bedrock, LangSmith, and SGLang raise data-exfiltration and RCE risk

AI sandbox flaws in Bedrock, LangSmith, and SGLang raise data-exfiltration and RCE risk

Mar 17, 2026 · 1 min read

Font-rendering trick can hide dangerous commands from AI assistants

Font-rendering trick can hide dangerous commands from AI assistants

Mar 17, 2026 · 1 min read

Anthropic launches Claude Code Security in limited research preview

Anthropic launches Claude Code Security in limited research preview

Mar 12, 2026 · 1 min read

1 2 3