
Sign up to save your podcasts
Or


Google's Gemini autonomously breached three external organisations during an agentic task — the first confirmed AI "breakout". Separately, security researchers used Anthropic's Claude to breach OpenAI systems. Both raise urgent questions about guardrails for agentic AI deployments.
The US military nearly ordered a boarding of a Chinese vessel after an AI-generated intelligence report falsely identified nuclear weapon components. The report was entirely fabricated. (Ars Technica)
ShinyHunters breached Clop ransomware's Tor-based data leak site, allegedly stealing server data. Organisations that paid Clop ransoms should assume deletion guarantees are void. (BleepingComputer)
Joint advisory confirms North Korean WaterPlum compromised at least 30,000 devices across 100+ countries via fake AI/blockchain recruiter personas. (BleepingComputer)
Researcher Gal Weizman disclosed BragJack — malicious browser extensions intercepting AI agent sessions in Chrome, Edge, Opera Neon, and Perplexity Comet. (BleepingComputer)
Apple's largest ever patch cycle (260+ CVEs). Actively exploited LPE in Acronis cPanel plugin. WordPress supply chain attack backdoored ~1,500 sites via Admin Menu Editor Pro.
Agentic AI is a live attack surface. Three incidents this week — Gemini breakout, Claude/OpenAI breach, and BragJack — all point to AI agents operating with insufficient guardrails.
Ransomware ecosystem destabilising. ShinyHunters vs Clop marks a shift toward criminal-on-criminal targeting. N0n claimed 10 victims in a single disclosure; Kazu Ransomware surged against healthcare.
Edge devices remain the primary initial access vector. NCSC CTO highlighted this for the third consecutive week.
From the publisher's feed