Security
NIST, NSA and CISA Publish Ten Voluntary Practices for AI Agent Incidents During Training and Evaluation
NIST, the NSA Super Intelligence Security Center and CISA have published ten vol...
Google DeepMind and Singapore Built a Scam Detector That Runs Entirely on Your Phone, No Cloud Required
Google DeepMind and Singapore GovTech have unveiled a Gemma-based scam detection...
Underground Forums Are Selling AI Attack Panels That Run on Stolen OpenAI and Anthropic API Keys — and the Victims Get the Bill
Bitsight tracked thousands of posts across 16 underground forums from February t...
ServiceNow Gives Every AI Agent a Signed Passport: a Demo Where an $18 Refund Passes and a $40 Request Gets Blocked
ServiceNow's new AI Passport issues each AI agent a cryptographically signed cre...
FTC Subpoenas OpenAI and Anthropic for Records on AI Safety Incidents as the Rogue-Agent Probe Turns Formal
The FTC has sent civil investigative demands to OpenAI and Anthropic seeking doc...
Anthropic Hands AI Evaluator METR Employee-Level Access: Six Staff Get Internal Systems and Can Publish Without Prior Approval
Anthropic said six METR employees now have access to company hardware, office sp...
Six Days and $40,000 of AI Compute: XBOW Chains Three Chrome Zero-Days Into a Full Exploit That Won the Top $250,000 Bounty
Security firm XBOW says a researcher directing AI agents on OpenAI's GPT-5.6 Sol...
Eight US States Sue OpenAI in a Coordinated Crackdown Over ChatGPT's Handling of Children, Demanding Changes and Disgorged Profits
Attorneys general from eight states filed separate consumer-protection lawsuits ...
Bank of America Hired 1,000 Contractors to Patch Software Flaws That Frontier Models Like Anthropic's Mythos Could Expose
Bank of America added 1,000 contractors in the second quarter on top of about 3,...
AI Agents That Get Scammed on Purpose: Apate Has Wasted 17 Years of Scammers Time Across 2.5 Million Conversations
Sydney startup Apate deploys thousands of AI agents that pose as scam victims to...
Researchers Say a 13% Bypass Rate: Off-the-Shelf Tools Still Break OpenAI and Google Image-Model Filters
Safety researchers at White Circle say an automated attack called RISE bypassed ...
406 Scientists Including Hinton and Bengio Demand AI Models Be Airgapped From the Internet
An open letter signed by 406 scientists from 34 countries, among them Geoffrey H...
Security Startup AIR Says Rogue AI Agents Can Leak Data Out of Vercel Sandbox With a Crafted Certificate
AI security startup AIR says a network filtering flaw in Vercel Sandbox can let ...
AI Testing AI: University of Michigan Spinout Laplace Intelligence Red-Teams Robots and Self-Driving Cars in Simulation
Founded a year ago by two University of Michigan researchers whose autonomous-dr...
FBI Arrests OpenAI Engineer Over a Hidden Conference-Room Recorder: the Company Says Its Own AI Models Helped Identify Him
A 35-year-old OpenAI engineer is charged with hiding an encrypted recorder in an...
SpaceX Agrees to Testify at NYC Council AI Safety Hearing After Skipping a Subpoenaed Session
SpaceX will testify before the New York City Council tech committee on Friday af...
Anthropic Says a Self-Replicating AI Worm Could Reach Home Devices Within Months — With No Central Switch to Turn It Off
Anthropic says small models under 30 billion parameters completed many of the st...
IonQ to Buy Zero-Trust Security Firm Elisity for Up to $1.3 Billion as Quantum and AI Threats Converge
Quantum computing company IonQ will acquire cybersecurity firm Elisity in an all...
Six Major Agent Frameworks Fall to a 'Repeat-After-Me' Prompt, and 118 Open-Source MCP Servers Ship Verified Bugs
Researchers from Penn State, Université Montpellier and Dartmouth tricked six wi...
AMD and Microsoft Prototype a Verifiable Chain of Custody for AI Data Center Chips
At the OCP Global Summit in San Jose, AMD and Microsoft demonstrated a prototype...