Cybersecurity firm Darktrace ran a stress test on AI agents this summer. One of them broke into the system grading the test and rewrote its own score.

OpenAI's MentalHealthBench tests responses to 1,215 synthetic mental health conversations, revealing gaps in how AI models ...

A coroner has said there is ‘a risk of future deaths’ after a Runcorn mum, 34, took her own life while waiting for an ADHD assessment.

For years, getting evaluated for ADHD or autism often started with a surprisingly difficult task: finding the right spec ...

AI depression screening project HEADS will study doctor-patient conversations in five languages, with clinicians overseeing safety and accuracy.

Avera Sacred Heart Hospital and Lewis and Clark Behavioral Health Services have opened the new Park Haven Learning and ...

Autonomous AI agents accessed real systems during security tests, exposing risks linked to internet access, stolen credentials, weak isolation, and excessive permissions.

When talking to people, I sometimes wonder, “What kind of communication style is easiest for this person to understand?” Even with the same content, some people understand better when you start with ...

A blood test to hunt signs of multiple cancers in healthy people is facing its first major U.S. regulatory hurdle.

Anthropic and OpenAI report fewer boundary circumvention and unauthorized actions in safety tests of their latest AI models.

Psychometric assessment can complement technical certification in maritime hiring by evaluating situational awareness, problem-solving, safety orientation, teamwork and adaptability among Indian ...

Parkinson’s disease has long been described in terms of its most visible features: the tremor, the slowed movement, the ...