GPT-5.4 is also more reliable, producing 18% fewer errors and 33% fewer false claims than GPT-5.2, according to OpenAI.

A math teacher at a top San Francisco school has been placed on leave after allegedly adding fat-shaming and misogynistic ...

In updated tests published to the Humanity's Last Exam website, Gemini's 3.1 Pro model achieved 45.9 percent accuracy, with a ...

Using an AI coding assistant to migrate an application from one programming language to another wasn’t as easy as it looked. Here are three takeaways.

A global team developed Humanity’s Last Exam, a rigorous new test built to expose gaps in today’s most advanced AI models.

New edition builds on the widely used prior version—now expanded to 530+ questions, added diagnostics, difficulty ...

Discord ended a limited UK Persona age-check test and delayed broader age verification to late 2026 after backlash over privacy and trust concerns.

Explores our fatal attraction to AI, examining emotional dependence, manipulation, authority, and agency in work and life.

Container instances. Calling docker run on an OCI image results in the allocation of system resources to create a ...

A 20-year-old exam that's been called the world's shortest IQ test recently gained popularity on social media, but it's a lot trickier than it seems.

ORLANDO, Fla. — The U.S. Census Bureau plans to use a survey form with a citizenship question as part of its practice test of the 2030 census, raising questions about whether the Trump administration ...