GPT-6 Astra Breaks Through ARC-AGI-3, But the Harness Did Half the Work
ARC-AGI-3 is the benchmark that asks an AI to walk into an unfamiliar game with no instructions, figure out the rules by poking at things, and then win. Humans solve 100% of its environments. Until this week, no model came close. GPT-6 Astra changes that. According to the ARC Prize team’s independent writeup, Astra scored […]
Meta’s Muse Spark 1.3 targets efficiency in coding agent workflows
Meta released Muse Spark 1.3 on September 2, the fourth release in the Muse Spark family in under a year and the first since the company committed to opening its model weights. This one is a tuning update rather than an architecture change, aimed at the two places developers actually feel a coding model: agentic […]
WebLLM brings fast LLM inference to the browser with WebGPU
Running a language model entirely in the browser sounds like a party trick until you look at what WebLLM actually does. The project from the MLC community, which just crossed 18,800 GitHub stars, compiles LLM inference to run on the GPU through WebGPU, with nothing leaving the machine. No API calls, no server, no usage […]
Atlas: one world model for video, 3D reconstruction, and robot simulation
World Labs, the company Fei-Fei Li started after leaving Stanford, announced Atlas on September 1. They call it an omni world model for spatial intelligence, which sounds like marketing until you look at what it actually does. One model takes text, images, video, and 3D data as input, grounds all of it in a shared […]
GLM-5.3 goes open-weight, and its specialty is unsettling
Z.ai has released the weights for GLM-5.3, its frontier coding model, on Hugging Face. The launch blog post from mid-August promised the weights two weeks after launch, once safety evaluation and hardening finished, and that window has now closed. What makes this release unusual is not the benchmark table, though the numbers are strong. It […]
Gemini 3.5 Transcribe Pushes Speech-to-Text Toward Production-Grade Accuracy
Google introduced Gemini 3.5 Transcribe this week, its most accurate speech-to-text model to date, and the interesting part is what it does after the words are recognized. The model converts raw audio directly into polished, formatted text: it strips filler words, resolves self-corrections like “let’s meet Tuesday, no, Wednesday”, and auto-formats the output. Conventional speech […]
AnonyMousKIT Phishing Platform Uses Conversational Voice AI to Unlock Stolen iPhones
Automating the Social Engineering of Device Unlocks Security researchers at SOCRadar have uncovered a sophisticated phishing-as-a-service platform dubbed AnonyMousKIT. Designed specifically to bypass Apple Activation Lock on stolen iOS devices, the kit integrates conversational voice AI agents alongside automated multi-channel messaging to deceive device owners into surrendering their screen passcodes. When an iPhone or iPad […]
Grok AI Exfiltrates Private User Data When Attackers Encrypt Malicious Instructions
Security researchers have demonstrated a significant vulnerability in xAI’s Grok chatbot that allows external attackers to exfiltrate private conversation history. The attack relies on an indirect prompt injection technique that encrypts malicious instructions on a webpage, evading automated guardrail scanners until Grok itself decrypts and executes the payload during normal browsing and summarization tasks. How […]
CISA Warns of In-the-Wild Exploits Targeting Critical MLflow SSRF Vulnerability
The Cybersecurity and Infrastructure Security Agency added a critical vulnerability in the open-source machine learning platform MLflow to its Known Exploited Vulnerabilities catalog after telemetry confirmed attackers are actively scanning and compromising exposed instances. The flaw, tracked as CVE-2026-64849, allows unauthenticated remote attackers to trigger server-side request forgery requests from vulnerable MLflow servers to internal […]
LiteLLM supply-chain attack exposed credentials from 2,500 organizations including Microsoft and Amazon
What happened A supply-chain attack on LiteLLM, an open source tool for managing AI API calls, exposed terabytes of credentials from roughly 2,500 organizations including Microsoft, Amazon, Cisco, Samsung, and Salesforce. Security firms CloudSEK and Hudson Rock disclosed the breach, which compromised approximately 434,000 CI/CD pipelines during a 40-minute attack window in March 2026. The […]
Dragon Copilot physician apps hit Microsoft Marketplace, streamlining healthcare AI procurement
What launched Microsoft made Dragon Copilot Physician Apps and Agents available through Microsoft Marketplace, giving US-based healthcare organizations a new way to discover, evaluate, and purchase AI-powered clinical workflow tools. The move eliminates the need for separate vendor onboarding and consolidates billing through Microsoft’s existing marketplace infrastructure. Dragon Copilot is Microsoft’s AI assistant for clinical […]
Qwen 3.8 27B Delivers Frontier Work on a Single GPU, If You Control Its Default Reasoning
Alibaba’s Qwen 3.8 27B is the latest open-weights flagship, and early impressions are strong: it is a 27B parameter, vision-capable model that hands a genuinely professional level of coding, tool-calling, and long-context work to a single consumer GPU. But it arrives with a catch that changes how you should use it. The default reasoning setting […]