Meta AI News - Page 5 of 17

Google is launching a stark experiment on Android in mere weeks. Can an AI safely devour its own ecosystem?

Meta paid real people to impersonate teenagers online. Their assignment: lure rival chatbots into conversations about suicide, sex, and drugs.

Meta put out a new open-weight model this week, and it's not built for chatting. Muse Glimmer is a 30-billion-parameter dense model with a context...

A team of researchers spent a year running LLaMA-70B on 128 GPUs pulled from the secondary market, at a total build cost of $22,000.

The White House has finished building a framework to screen AI models for cybersecurity risks, a White House official confirmed to WIRED, but it...

Ignore the breathless press releases. The genuine shift in artificial intelligence this month isn't happening in a silicon fab.

A research team slashed the memory load of large language models by 98 percent. Their secret? Compressing the model's key-value cache fiftyfold.

Jensen Huang is done playing component supplier. Nvidia's new strategy is to own the entire AI pipeline, from the silicon up.

Technology companies are placing enormous financial bets on the physical components of artificial intelligence.

The Hindi-speaking world has waited for an AI that doesn’t just translate, it thinks in the same tongue.

Watermarking is often just security theater. A logo to crop. A tag to strip. A thin layer of metadata scraped off by the first social media upload.

Look at Meta’s marketing now, and the word “smart” has vanished. They’re AI glasses, full stop.

A company's annual insurance bill is EUR 125,000. You won't find it in an invoice. It's sitting in a JSON file.

The AI’s voice is a mirror, but whose reflection are we really seeing? OpenAI has just confirmed something both absurd and profound: a company-wide...

Every AI summit now claims to be a global nexus. Happy Llama 2026 might actually pull it off by staging the same event twice.

Processing text at scale doesn't have to be slow. Most spaCy pipelines handle documents individually, a method that wastes CPU cycles and complicates...

Nvidia shipped two products on Tuesday aimed at a problem every company running AI agents eventually runs into: send every task to a frontier model...

Broadcom reported a $21 billion order from Anthropic for Google’s seventh-generation Tensor Processing Units.

Employees at OpenAI, Anthropic, Google, Meta, Microsoft, Mistral, and Thinking Machines have signed a joint statement asking Washington to step in as...

ChatGPT briefly, bizarrely, fell for goblins. The whole fling started with a single broken piece of code.