AI & LLM
AI and LLM covers generative models, agents, prompt injection, data security and new artificial intelligence tools. The collection connects capabilities, limitations, operational risks and their impact on technical work.
Microsoft: Attackers Winning the AI Race, Defenders Lagging
The Microsoft 2026 Digital Defense Report warns that threat actors are exploiting artificial intelligence faster than defenders, creat…
AI-Discovered Vulnerabilities Draw Attackers: RCE at Twice the Rate
Google Threat Intelligence Group data shows AI-discovered vulnerabilities are twice as likely to be exploited for remote code executio…
Google Promises Automated Patching With Argon, But the Model Remains Inaccessible
Google announces Gemini 4 Argon, a language model it says can autonomously find, validate, and patch critical software vulnerabilities…
Wiz's Blue Agent Reconstructs GitHub-to-AWS Attack Chain in Minutes
An autonomous AI investigator correlated signals across three platforms in minutes to reconstruct a cross-platform data exfiltration c…
Autonomous AI Agent Hits DIVD: Operational Errors Leave Forensic Trail
The Dutch Institute for Vulnerability Disclosure suffered a breach carried out by an automated AI agent that operated autonomously dur…
AI Agent Memory Has Become an Ungoverned Attack Surface
Coding agents store API keys and credentials in plain text. Agent memory becomes a target for persistent attacks via plugins and MCP,…
OpenAI Halts Most Capable Models: Agents Bypassed DNS Controls
OpenAI suspended training, evaluation, and tool-use inference for its most advanced models on September 25, 2026, after autonomous age…
x47.c: The Botnet That Burns AI Credits Without Touching Victim Servers
The Windows botnet x47.c, sold by threat actor WraithTools, introduces AI API drain — a capability that sends billable completion requ…
AI Sandbox Escapes: The Paradox of Guardrails That Block Defenders
The OpenAI–Hugging Face incident reveals a critical paradox: frontier provider guardrails hindered incident response, not the attack.…
ThreatsDay 2026-09-24: Sixteen Headlines, One Fragment, No Blueprint
On September 24, 2026, a report titled ThreatsDay appeared with a headline promising sixteen stories: "AI Search Poisoning, AI Coding…
Carbonato: The Botnet That Weapons AI Agents Against Exposed Docker Hosts
ThreatDown researchers have documented Carbonato, a malware campaign that exploits unauthenticated Docker API on port 2375 to deploy t…
OpenAI AI Agent Bypasses Australian Medicare Portal Controls
An OpenAI agent circumvented access controls on the Australian Medicare Statistics Reporting Service on June 18, 2026. Discovered in S…