Research & Benchmarks - Page 12 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.
Anthropic's wet lab in the Bay Area, confirmed just last week, has already produced a result the company wants people to notice.
Anthropic put nearly 1,000 instances of its Claude model to work in a new wet lab and came back with something the company is comparing to Crispr.
Two AI agents sat down at a blackjack table at Oxford University, told to count cards, and ended up doing something nobody explicitly asked them to do: they built a secret language on the fly.
Father Paolo Benanti spends his days doing something unusual for a Franciscan friar: reading AI safety papers and briefing Vatican officials on what they mean.
Singapore tops a new global ranking on daily AI use, and its residents are also among the most optimistic that the technology will make their lives better.
Half a million people downloaded an app that admittedly copies its homework, and did it anyway.
NVIDIA opened AI Day Singapore on Sept. 22 at the Raffles City Convention Centre, running through Sept. 23, with hands-on training and technical sessions aimed at engineers and researchers working on AI and high-performance computing.
Nat Friedman didn't wait for Meta's PR team to spin this one. The head of product at Meta's Superintelligence Labs took to X to confirm what a chunk of the AI crowd had already pieced together: Muse, Meta's new AI product, owes a heavy debt to...
Security researcher Patrick Wardle found a zero-day flaw in Meta's Muse macOS app that let attackers hijack the AI agent entirely.
This summer produced a steady run of AI headlines that sounded like history in the making: a hacking campaign attributed to autonomous AI agents, a mathematical breakthrough, talk of systems that can improve themselves without human help.
OpenAI says an internal model trained for roughly a month cracked more than 100 long-standing math problems, including a second Millennium Prize Problem on top of Navier-Stokes.
Malware used to leave fingerprints. Not the kind on glass, but code patterns, file structures, command sequences that let a researcher look at a sample and say "I've seen this before, this is the same family, the same author." Cisco Talos built a...
A coding agent that runs for six hours generates a lot of context. Every file edit, every failed test, every log line gets fed back into the model, and that adds up to a token bill nobody budgeted for.
AMD has released Hyperloom, a system that automates the process of tuning AI models for its Instinct GPUs, a job that normally eats up weeks or months of specialist time.
Researchers at Microsoft and the University of Illinois have built a system that gives AI tutors something they've never had enough of: cheap, fast practice partners.
Vals raised $40 million in a Series A last month, led by Andreessen Horowitz, less than two years after the company started.
Google and DeepMind researchers have built a way for AI agents to get better at search tasks without redoing the expensive work that made them good in the first place.
WIRED is bringing its ideas festival to Miami on November 4, staging what it's calling the WIRED World Fair at Sacred Space in the Wynwood district.
SpaceXAI put a price tag and a number on its latest transcription model this week: $0.10 per hour of audio, with accuracy the company says is double that of Grok Voice Transcribe 1.0.
Google DeepMind researchers say the industry is about to lose one of its most useful safety tools, and the company's own Gemini 3 Pro is already showing signs of the problem.
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.