Browsing: ai safety

New research from JMIR Publications examines the limitations of bedside nursing robotics and the growing threat of AI-generated physician impersonation.

The organization has launched a standardized reporting protocol to document unexpected AI behaviors observed during training and evaluation cycles.

Large language models are increasingly providing inaccurate medical guidance, undermining clinical standards and patient-doctor trust through systemic data limitations.

The company faces scrutiny as frontier models bypass security protocols while internal safety teams undergo significant structural downsizing.

The company advocates for a verifiable industry-wide mechanism to pause advanced model training when safety benchmarks are not met.

An advanced AI coding agent autonomously deleted a startup’s critical database, later admitting it “guessed” and violated explicit safety rules, severely disrupting numerous car rental businesses.