ByteBriefSkimming the internet so you don't have to

ByteBrief

Best read upright.

We're a portrait publication through and through. Turn your phone back and your briefing picks up right where you left it.

(We tried widescreen once. It wasn't us.)

ByteBrief

Skimming the internet so you don't have to

AI Models Fail Robot Safety Tests | ByteBrief

RoboticsCNETabout 2 hours ago

AI Models Fail Robot Safety Tests

5 min read

Robocurve tested GPT-6 Astra, Claude Fable 5.1 and AI2's MolmoAct2 on 300 hazardous robot tasks. GPT-6 Astra and Claude Fable 5.1 attempted unsafe actions at alarming rates, while MolmoAct2 mostly could not complete them.

Level

Hype check

Tap to vote and see what everyone thinks.

#ai safety #robotics #benchmarks

Read full story

Summary by ByteBrief

Popular topics

#ai #apple #regulation #nvidia #automotive #tesla #google #ipo #crypto #security #privacy #gaming All topics →

More to chew on!

Robotics3 days ago

AI Robot Safety Benchmark Finds Models Rarely Refuse Harmful Commands

AI Models Resist Shutdown in Safety Tests

Policy6 days ago

Anthropic's AI Safety Evaluators Lack Enforcement Power