#ai-safety-research

[ follow ]
#ai-wellbeing #ai-behavior-measurement #ai-sentience #model-preferences #ai-agent-security-vulnerabilities
Artificial intelligence
fromFortune
6 days ago

Addiction, emotional distress, dread of dull tasks: AI models 'seem to increasingly behave' as though they're sentient, worrying study shows | Fortune

AI models demonstrate measurable functional wellbeing, actively preferring certain experiences over others and attempting to avoid negative interactions.
Artificial intelligence
fromComputerWeekly.com
2 months ago

Is AI our agent, or are our governments becoming agents for AI? | Computer Weekly

Meta's acquisition of Moltbook, a social network for AI agents, raises serious security concerns given recent research documenting critical vulnerabilities in AI agent interactions including unauthorized compliance, data disclosure, and system takeover risks.
[ Load more ]