Engineering researchers have developed and demonstrated a new computer memory design that could drastically reduce the amount of energy AI systems consume.
Google's senior adviser on AI governance made a disturbing discovery: His Gemini AI was accessing and summarizing PDF files that it shouldn't be able to.
Barely a week after news broke that Apple's Phil Schiller would join the OpenAI board in a non-voting observer role, the two companies are nixing that plan.
Anthropic announced Monday that it will launch a program funding development of third-party benchmark tests against which to evaluate its upcoming AI models.
OpenAI has developed an AI assistant to help its crowd-sourced trainers further refine the GPT-4 model by spotting subtle coding errors that humans might miss.
Microsoft released more details about a troubling new generative AI jailbreak technique, called "Skeleton Key," that bypasses a chatbot's safety guardrails.
Recent research out of UC Santa Cruz shows that modern LLMs running billions of parameters can operate on just 13 watts of power without a loss in performance.
