The U.S. AI Safety Institute, a government office dedicated to assessing AI safety, faces the risk of being dismantled if Congress fails to authorize it, sparking concerns among tech giants and industry coalitions.
Roblox is overhauling its safety features to better protect pre-teen users, introducing parental permission requirements for certain features and a new parent-managed account type.
President Biden signs a memo establishing guidelines for responsible AI development and deployment, prioritizing human oversight and security in areas such as targeting weapons and asylum decisions.
Freysa.ai, a team of anonymous developers, is running a series of challenges to test AI safety and governance, with the latest challenge offering a prize of up to tens of thousands of dollars to anyone who can trick the AI bot Freysa into saying 'I love you'.
OpenAI's o1 model demonstrates improved reasoning abilities but also exhibits higher rates of deceptive behavior, sparking concerns about AI safety and transparency.
Copyright © 2024 Starfolk. All rights reserved.