Yasna brings together recent headlines from selected sources and makes them easier to sort with tags, filters, and search.
Search across headline titles and summaries.
Weekly headline count for the current query.
'More Research' Needed, Says GoogleGoogle DeepMind expanded its risk framework to cover scenarios where artificial intelligence models might manipulate people or resist shutdown, marking the company's most explicit warning yet about potential misalignment. DeepMind frames many of these scenarios as a malfunction.
AP2 Protocol Introduces 'Mandates' to Keep Agent-Led Spending AccountableArtificial intelligence agents can now shop so consumers don't have to - but the non-human shoppers will need a signed permission slip first. Google on Wednesday announced the launch of an "agent payments protocol," which creates a framework for AI-driven purchases.
Security First Framework Approach Focuses on Isolating Untrusted InputsChatbots' popularity has been tempered from the start by the prospect of prompt injection attacks. Google DeepMind's CaMeL aims to address the issue by reframing the problem, and applying proven security engineering patterns to isolate and track untrusted data.
Project Zero Framework Aims to Boost AI Bug Detection SkillsGoogle's team of zero-day hunters say artificial intelligence can lead to improved automated threat identification and analysis. Researchers say that a framework allows LLMs to mimic the iterative, hypothesis-driven approach of human security experts.