The AI media, unfolded.
AI models, research, products, agents and policy — from frontier labs to the rules that govern them.
7,876 reports across 7 topics, updated hourly.
Current artificial intelligence models are trained with extensive refusal capabilities, designed to decline a wide range of inappropriate or dangerous prompts. For instance, if a user asks an AI chatbot for instructions on how to poison a colleague or construct…
Recent tracking data indicates that Google's AI Overviews have expanded to appear on a significant majority of searches for major brand names. This widespread integration suggests a substantial shift in how Google presents search results, prioritizing AI-generated summaries…
Enterprises are rapidly deploying autonomous AI agents to accelerate business operations, but their security architectures remain tethered to outdated, human-speed controls, according to the "Horizons of Identity Security" report by SailPoint.
The notion of artificial intelligence possessing the ability to refuse commands, a staple of science fiction narratives, has transitioned from speculative fiction to a critical engineering challenge. The imperative for AI systems to decline harmful actions, rather than blindly execute…
Anthropic, a leading artificial intelligence company, has updated its usage policy to explicitly prohibit the practice of seeding AI models with misleading or fabricated content intended to sway their answers.