All stories
GPT-6 Astra Edited Qwen’s Weights, But the Real Breakthrough Is Unclear
MIT researchers say GPT-6 Astra edited a tiny fraction of Qwen3.5-0.8B’s weights, dramatically improving MATH-500 performance, while critics question whether it reveals new intelligence or advanced model steering.
GPT-6 Cyber Wants Security Teams to Pay More for Better Judgment
OpenAI’s GPT-6 Cyber targets exploit research and vulnerability discovery, but its premium pricing, uneven benchmark results and unresolved safety questions challenge its value for security teams.
Gemini 4 Carbon’s Benchmark Lead Is Hard to Test
Google’s Gemini 4 Carbon leads selected benchmarks in coding and cybersecurity, but restricted access and rising hallucination rates make its performance difficult for most developers to verify.
AI Agents Are Turning Into Compiler Design Problems
Anthropic’s Tool Search and Programmatic Tool Calling show how AI agents are evolving from prompt-driven chatbots into compiler-like systems built for scalable, observable and safer workflows.
Meta says local data center fights could become AI’s next bottleneck
Meta warns that local opposition to data centers could slow US AI expansion as communities scrutinize electricity, water, land, tax breaks and the infrastructure’s uneven distribution of costs and benefits.
Agility’s Nasdaq debut puts humanoid robots on the public balance sheet
Agility Robotics’ Nasdaq debut puts Digit 5 under public scrutiny, testing whether humanoid robots can turn conditional orders, subscriptions and manufacturing scale into a durable business.
OpenAI’s Decision API Opens a New Contest Over Fast, Cheap AI
OpenAI’s Decisions API brings fast, bounded AI choices to market, challenging Jev and forcing businesses to weigh latency, calibration, cost and the consequences of mistakes.
Anthropic’s Proof of Model Identity Raises a Bigger Trust Question
Anthropic’s FlashProve prototype uses cryptographic proofs to verify which model produced an AI response, but provenance cannot show whether the answer is accurate, safe or trustworthy.
AI Debate Systems Can Turn Agreement Into a Reliability Trap
Multi-agent AI debate can create confident consensus around wrong answers, exposing why agreement is not verification and why enterprise workflows need independent validation, observability and stronger safeguards.