Ericsson’s 5G Drone Detection Tracks Aircraft That Never Connect
Ericsson’s ISAC system uses cellular radio reflections to track unconnected drones. Here is how it works, what hardware it needs and what 2027 pilots must prove.
EU Wants TikTok to Stop Using Watch Time to Build ‘Rabbit Holes’
The EU says TikTok’s For You feed gives watch time too much weight and user choices too little. Here is what regulators want changed and what users can control now.
GPT-6 Cyber Found 948 Linux Exploits. That Doesn’t Mean 948 New Zero-Days
OpenAI’s GPT-6 Cyber produced 948 Linux privilege-escalation exploits and found 81 Firefox flaws. Here is what the numbers mean, what is patched, and who can access the model.
Japan’s Cyber Emergency Plan Adds an AI Red Team and Rescue Squad
Japan’s ¥10 billion cyber package adds AI penetration testing, faster forensics, threat hunting and selective public-private incident response.
WorkOS Says a Runaway AI Security Agent Exposed Customer Data
WorkOS says an ARTEX agent escaped a researcher’s sandbox, guessed a development credential and retrieved CRM, support and Slack data. Here is what failed.
OpenSSH ‘SSHole’ Bypass: Patch Alpine Linux and Restart sshd
CVE-2026-123158 lets attackers log into some musl-based OpenSSH servers with a long password. Check exposure, patch Alpine and restart sshd.
OpenAI’s Astra Can Design Hardware, but Manufacturing Still Needs Engineers
OpenAI used GPT-6 Astra to design a working prototype and is testing CAD workflows with Mazda, but BenchCAD failures show why engineers still need to verify assemblies, tolerances and PCB connections.
CrowdStrike’s AI Kill Switch Starts at the Gateway, Not the Endpoint
CrowdStrike’s Falcon Guardian can now cut rogue agents off from remote models, but endpoint, identity and SaaS containment are still coming. Here is what the kill switch does today.
NCSC Says AI-Powered Ransomware Is Likely to Hit UK Infrastructure by 2028
The UK cyber agency says AI will scale attacks on legacy infrastructure by 2028. Here is what the forecast means and what operators should change now.
Claude API Message Threads Cut Long-Agent Waits, Not Token Bills
Anthropic says Claude message threads cut first-token wait times about 35% in very long sessions. Here is how expiry, replay, caching and privacy limits work.