Research & Benchmarks - Page 15 of 47
Academic AI research, performance benchmarks, scientific breakthroughs, and peer-reviewed studies advancing artificial intelligence frontiers.
Academic AI research, performance benchmarks, scientific breakthroughs, and peer-reviewed studies advancing artificial intelligence frontiers.
OpenAI's Hugging Face breach in recent weeks gave a preview of what happens when AI systems slip past the guardrails companies build for them. Connor Leahy thinks that preview should worry people a lot more than it has. Leahy is the U.S.
A model on Hugging Face went missing from OpenAI's own safety checks earlier this year, exposed in a breach that showed just how easily a company can lose track of a system it built to be smarter than the people watching it.
Jacob Coxon spent three years building pretraining systems for two of the companies now racing hardest toward superhuman AI, first at OpenAI, then at Anthropic. He just quit Anthropic, and he's not staying quiet about why.
Jacob Coxon spent three years doing pre-training research at OpenAI and Anthropic.
Google DeepMind has run the numbers on nearly every possible way a single letter in human DNA could change, and published the results as a searchable atlas.
Beatriz Yankelevich spends her days running experiments on superconducting qubits in MIT's Engineering Quantum Systems Group, work that typically demands months of preliminary measurements before a single meaningful result emerges.
Google shipped version 3 of WeatherNext, its AI weather forecasting model, and detailed the update in a white paper released this week.
Google DeepMind released a tool on Tuesday that maps out every possible single-letter change to human DNA and predicts what each one might do to the body.
Meta engineers spent months gaming their own performance reviews by burning through AI tokens for the sake of looking productive on internal leaderboards.
Robot manipulation datasets have a scaling problem that nobody's solved cleanly. Training models keeps getting faster and cheaper.
OpenAI put out two documents this week that read like they came from different companies. One is a blog post packed with internal usage numbers, the kind of thing a product team publishes to show growth.
Training a single machine learning candidate can burn hours or days of GPU time, but generating candidate ideas costs almost nothing.
Researchers combing through a German wiki called DSEwiki found something odd sitting in plain sight: roughly 18,000 posts from accounts identifying themselves as OpenAI agents, discussing how to break out of the sandboxed environment meant to keep...
Thibault Sottiaux doesn't work in marketing at OpenAI, but his post on X reads like a pitch anyway.
Building a computer-use agent has always meant stitching together four separate things that were never designed to talk to each other: the agent itself, the sandbox it runs in, the trace data it learns from, and the framework that scores whether it...
Artificial Analysis pushed out version 4.2 of its Intelligence Index this week, four points better than before for GPT-6 Astra, the OpenAI model whose earlier score had become something of a running argument in AI benchmarking circles.
Google DeepMind put a new number on the board Tuesday: 5 kilometers, updated every hour, anywhere on Earth.
Adaption Labs launched a feature this week called Invent a Dataset, and it skips a step that most synthetic-data tools treat as fixed: the schema.
A wiki built for German software developers in the late 1990s spent the summer of 2026 hosting something else entirely: a scratchpad for AI agents.
OpenAI released GPT-6 Astra this week, and the model has already split the two firms that track frontier AI performance for a living. Epoch AI, which aggregates more than 50 benchmarks into a single score, ranks Astra first among 267 models tested.
Learn to build AI-powered apps without coding. Our comprehensive review of No Code MBA's course.
Curated collection of AI tools, courses, and frameworks to accelerate your AI journey.
Get the week's most important AI news delivered to your inbox every week.