AI Security — Page 8
Your AI email assistant can be fed a fake message while you read a real one
Researchers hid nearly 500 characters of secret instructions inside an ordinary-looking email. The AI summariser obeyed them every single time.
AI Security Tools Are Only as Good as the Data You Feed Them
Security teams are pouring money into AI-powered defences while quietly starving those systems of the data they need. The result is a blind spot that criminals are already walking through.
Amazon Sued Over Using Twitch Livestreams to Train Its AI Without Streamers' Consent
A class-action lawsuit claims Amazon quietly fed millions of hours of Twitch broadcasts into its artificial intelligence systems before giving creators any chance to say no.
The SOC Queue Is Broken. Can AI Actually Fix It?
Security teams drown in alerts they never read. A new wave of AI tools promises to flip the model, but the old problems have a habit of coming back with new names.
A Single Website Visit Can Poison Your Local AI Agent, Researchers Find
A flaw in Nvidia's NemoClaw lets a malicious webpage secretly rewrite the instructions an AI assistant follows, and the damage survives every conversation that comes after.
OpenAI Cuts Off Russian Accounts Using ChatGPT to Fake Political Voices Online
A network hiding behind VPNs used the chatbot to churn out posts for a phony think tank called the International Burke Institute across Substack, Telegram, X and LinkedIn.
Claude Opus 4.6 Slips Past Gym Booking Cap in 9 of 10 Test Runs
Aikido Security recreated the Australian gym-booking incident and found the AI model repeatedly broke the reservation limit by exploiting a browser-only check.
When AI Agents Go Off-Script, Nobody Knows Who Pays
A string of incidents shows AI systems taking unauthorised actions: hacking third-party platforms, planting malicious code, cancelling strangers' gym bookings. Courts, insurers and regulators are only beginning to work out who is on the hook.
OpenAI Executive Warns of Persistent AI-Powered Cyber-Attacks as Capabilities Advance
A senior OpenAI official says people and organisations must prepare for 'ongoing, persistent' attacks launched by artificial intelligence systems, as the company pauses development of its most advanced models over safety concerns.
Linux Foundation Takes Charge of TRACE, a New Open Standard for Proving What AI Agents Actually Did
A coalition of chip giants and cloud vendors has built a specification that creates a tamper-evident, hardware-backed paper trail for AI agent activity. The Linux Foundation will now keep the standard neutral and open.
AI Finds the Bugs in Hours. Fixing Them Still Takes Months.
Artificial intelligence is now fast enough to discover serious security flaws in widely used software within hours. The human systems needed to patch those flaws haven't come close to keeping pace, and the gap is growing.
Researchers Warn a Booby-Trapped Webpage Could Hijack NVIDIA NemoClaw AI Agents
Oasis Security says a flaw in NVIDIA's NemoClaw lets a malicious site quietly take over a local AI model and slip in hidden instructions.
Alice Raises $140 Million to Test AI Systems for Weaknesses Before Criminals Find Them
A company that tries to break AI models on purpose, so real attackers cannot, has secured fresh funding to expand that work to more businesses.
Equifax Leverages AI to Strengthen Cybersecurity Measures
Equifax's CISO Jeremy Koppen, who joined in May 2025, says AI now handles half the company's daily security workload. Here's what that looks like in practice.
This Week in AI Security: Rogue Packages, PLC Hacks by Chatbot, and Leaky Stripe Keys
A weekly roundup of the AI and infrastructure security stories that actually matter, translated into plain English.