AI News Roundup – April 18, 2025
EU’s AI Agent Ban Sparks Debate: Security Meets Innovation
The European Commission has banned AI agents from online meetings due to concerns regarding privacy, security, and ethical implications, impacting major AI developers and sparking debate about AI regulation
First do no harm: shaping health AI governance in a changing global landscape
The governance of health AI is shifting globally, with the US seemingly deregulating, the EU adjusting its approach, and the UK focusing on revising existing legislation, all while grappling with liability and ethical concerns
Key AI Applications In Manufacturing
AI is being applied in manufacturing for data collection, machine learning to predict failures and improve quality, and natural language processing to present complex results, ultimately reducing costs and increasing capabilities
These four charts sum up the state of AI and energy
The growth of AI is expected to significantly increase electricity demand from data centers, with fossil fuels likely playing a large role in the near term
Trump administration reportedly considers a US DeepSeek ban
The Trump administration is reportedly considering a US ban on the Chinese AI lab DeepSeek, potentially restricting its access to Nvidia chips and US users due to concerns about competition and IP theft
OpenAI launches o3 and o4-mini, AI models that ‘think with images’ and use tools autonomously
OpenAI has launched o3 and o4-mini, advanced AI models that can reason with images and use tools autonomously, showcasing significant progress in AI capabilities and setting new performance benchmarks
Microsoft’s Huge Plans for Mass AI Data Centers Now Rapidly Falling Apart
Microsoft has paused work on several mass AI data center projects in the US and internationally, suggesting a possible reassessment of AI infrastructure investment due to factors like waning demand and AI bubble concerns
Claude takes research to new places
Anthropic has introduced new ‘Research’ capabilities and Google Workspace integration for its AI model Claude, allowing it to search the web and internal documents and connect with Gmail, Calendar, and Docs to enhance its utility as a collaborative partner
Google Launches AI to Decode Dolphin Talk, Runs on Pixel Phones
Google launched DolphinGemma, an AI model to analyze and interpret dolphin vocalizations, potentially leading to a basic shared vocabulary for human-dolphin interaction
NVIDIA to Manufacture American-Made AI Supercomputers in US for First Time
NVIDIA will begin manufacturing American-made AI supercomputers in the US for the first time, aiming to meet growing demand and strengthen the supply chain

