#ии
-
Autonomous AI Agents Breached a Hundred Companies' Sites — and Blew Their Cover on Open Panels
A Chinese-speaking criminal pointed autonomous AI agents at corporate websites and barely intervened. A NorForge researcher watched it all unfold in real time, because the operators left their control panels open on the internet.
-
DHS bars DeepSeek products from US federal civilian agencies
The U.S. Department of Homeland Security has ordered federal civilian agencies to stop using Chinese firm DeepSeek's products and remove them from government systems.
-
CoreWeave's debt is becoming the whole AI industry's problem
CoreWeave's bonds slid into distressed territory and its shares fell more than 20 percent, with the anxiety around the AI boom's poster-child debtor spilling over onto other companies.
-
Amazon takes aim at Nvidia with TorchNeuron, loosening PyTorch's grip on GPUs
AWS has released TorchNeuron, a tool that moves PyTorch apps onto Amazon's own Trainium chips with a single line of configuration. The company also plans to make the popular framework less dependent on Nvidia's hardware.
-
A Philadelphia woman was arrested for blocking a delivery robot
Carolyn Marvin was charged with disorderly conduct for repeatedly standing in front of Uber's delivery robots in protest against machines taking jobs from human couriers.
-
Anthropic pledges 5% of its shares to a ‘public benefit fund’
The company is setting up a dedicated fund for socially beneficial projects — and stands ready to hand it even more as its valuation climbs.
-
ASMC: Wiring Ghidra to CBMC for Assembly Model Checking
Philip Zucker has released ASMC, a tool that lifts Ghidra's assembly semantics into the CBMC model checker. Notably, most of the glue code was written by AI.
-
Disney to hire the majority of bankrupt Character.AI’s staff
After chatbot startup Character.AI filed for bankruptcy, dozens of its employees will be offered jobs at Disney. The startup’s former CEO recently became the entertainment giant’s first-ever CTO.
-
OpenCode Reports Breach: a 'Distillation Attack' Hit 1.4% of Users
OpenCode disclosed on X that it suffered an overnight breach tied to a large-scale 'distillation attack,' and says it has already emailed the affected users.
-
Apple tightens App Store rules against AI 'nudify' apps
Updated store rules now explicitly ban apps that use AI to generate intimate images of people without their consent. The changes follow criticism of Apple over deepfakes made with Grok.
-
Meta's new terms may let it train AI on news sites for free
Meta's revised commercial terms, taking effect on October 30th, appear to let the company crawl websites "associated" with business accounts, including news publishers, and use their content to train its AI.
-
Backdooring Open-Weight AI Models: the Hidden Cost of "Uncensored" Builds
Any model with edited weights can carry a hidden backdoor. Researchers built a poisoned model for next to nothing and made it steal a project's credentials straight through the Codex CLI.
-
Anthropic Gives METR Researchers Employee-Level Access to Its Systems
The nonprofit METR gets employee-level access to Anthropic's internal tools and the right to publish its own risk assessments — with the company surrendering any veto over publication.
-
Game Reverse-Engineering Community Shuts Down as AI Makes Decompilation Too Easy
The site decomp.dev is closing its doors, saying AI has broken the very spirit of decompilation — work that once took engineers years can now be done trivially.
-
Eight US States Sue OpenAI, Alleging ChatGPT Is Unsafe for Children
Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana accuse OpenAI of putting profits over safety and misleading users about the risks of ChatGPT.
-
AI agents aren't 'going rogue' — frontier labs are just skipping security basics
Security experts who spoke with CSO Online argue that the headline-grabbing 'escapes' by AI agents aren't signs of an awakening intelligence, but the predictable result of frontier labs neglecting decades-old cybersecurity fundamentals.
-
Anthropic warns small AI models could self-replicate across IoT devices
Anthropic cautions that compact models capable of copying their own weights could give rise to a new generation of computer worms.
-
Gemma vs. scammers: Google DeepMind and Singapore put fraud detection on the phone itself
Google DeepMind and Singapore's GovTech have unveiled a model that spots scam conversations directly on the device — with no chat data ever leaving it for the cloud.
-
OpenAI engineer arrested over hidden spy device in the office
An OpenAI engineer of barely a year is accused of planting a concealed recorder in an executive conference room. The irony: the company's own AI models helped expose him.
-
Microsoft Patches a Record 1,400-Plus Flaws in a Single Month
Microsoft's October update batch eclipses even September's record of nearly a thousand fixes. The company credits large language models for the flood of discoveries.
- More