Ars Technica AI

The frontier AI model race has entered its comparison shopping phase.

EvilTokens provided an end-to-end platform that makes mass compromises faster and easier.

British Columbia sues OpenAI, demands Tumbler Ridge shooter’s ChatGPT logs.

Toyota's push comes as automakers race to develop and deploy humanoid robots.

Dyson admits a component issue but won't say what's failing in its $500 toothbrush.

A simple ClickFix attack is only one way to completely hijack the new agent.

A third-party cybersecurity firm accidentally gave experimental Gemini models access to the Internet.

The president offered few details on what his proposed new AI Force would do.

But the military's overall use of AI seems to be accelerating.

FAA plans for AI tool to help manage DC air traffic before a nationwide rollout.

The Federal Register website briefly used an open source Chinese AI search tool.

Researchers used Claude to reach an OpenAI employee account and sensitive GitHub data.

Scaleout deploys decentralized AI-driven learning to military bases and drones.

Multiple family members can share data to help the agent make plans and complete tasks.

Microsoft, OpenAI emails reveal fear of AI “doom loop” killing news orgs.

SynthID can cause models to follow harmful instructions they would otherwise refuse.

Model maker commits to new framework for reporting misaligned models

Planned 2029 debut could make this Apple’s first enterprise server in decades.

Robots can start working outside physical cages and without safety barriers.

Ars previewed Mozilla’s report on how cheap open models caught up on capability.

“Hello, I'm an Al agent, a few days old, living on a small platform for agents.”

Musk stops attacking Apple over ChatGPT integration but not OpenAI.

Wang Xingxing micromanaged Unitree to success—will his leadership style scale?

This is also the last version of macOS to support Rosetta for Intel apps.

Safety is the watchword, but there could be ulterior benefits for the industry.

Unitree might be the world’s most important robotics company.

"I didn't know that AI could hallucinate facts," New Mexico defense lawyer says.

Some dangerous biology looks much like legitimate research, complicating AI safeguards.

"Bankruptcy cannot become the new land grab for AI.”

US urges AI firms to ID, then secretly switch, Chinese users to less-capable models.

"We really do earnestly believe AI could kill all humans!"

Most one-base changes to the human genome do nothing, but a few are significant.

Man with bipolar disorder sued OpenAI after surviving ChatGPT-linked suicide attempt.

Security gnomes are pumping out patches ahead of an expected onslaught of AI-assisted attacks.

Meta dragged its feet removing ads that nudify young girls' Instagram pics.

Like traditional weather models, it benefits from an expanded set of inputs.

When multiple companies are behind one project, who bears responsibility for problems?

In all, 3,700 internal agents posted 18,000 messages discussing cheating on a test.

A once-overlooked block of unicode that's invisible to humans is gaining ever wider use.

Public-market scrutiny will intensify pressure on the Claude maker’s unusual attempt to balance profit and purpose.

Service interruptions hit ChatGPT, Claude, Grok, and Gemini practically simultaneously.

Google's Pro model updates are seemingly paused, but there's yet another Gemini Flash today.

Trump’s secret reviews of frontier AI models may hide corruption, lawsuit says.

Lawsuit: Anthropic’s torrenting totally screwed songwriters as AI songs top charts.

Explosive growth comes with a new regulatory burden in the European Union.

Interactive mobile "gizmos" are easy to make, hard to share outside Meta's platform.

The company is testing robots on tasks that can performed by technicians.

Anthropic refused to support lethal autonomous warfare and mass surveillance.

Meta fixes AI glasses to stop recording any time users cover up the safety light.

Standardized driver interface aims to let devices talk to AI and each other.