The afternoon briefing.

AGI Debates. Marc Andreessen, a prominent venture capitalist, asserted on a podcast that frontier AI models have already achieved Artificial General Intelligence (AGI), a claim that continues to spark debate among tech leaders about the shifting boundaries of AI capabilities. This bold statement underscores the rapid advancements in large language models and their increasing sophistication. The discussion highlights the ongoing challenge of defining and recognizing true AGI.

Policy & Military AI. The Trump administration has issued a memo directing the military to integrate the "most advanced AI in the world" into its operations, while also preventing companies from altering these models without prior approval. This move signals a significant push for AI adoption in defense. Concurrently, a US House draft bill aims to prohibit individual states from enacting their own AI regulations, seeking to establish a unified national policy framework.

Infrastructure & Regulation. New York is moving closer to becoming the first US state to implement a temporary moratorium on large data centers, specifically targeting "hyperscale" facilities over 20MW, reflecting growing concerns over energy consumption and local impact. This legislative action highlights the increasing tension between the demand for AI infrastructure and environmental or community considerations. The debate underscores the need for balanced growth in the AI sector.

AI Security & Development. Meta confirmed that thousands of Instagram accounts were compromised through the abuse of its AI chatbot, revealing vulnerabilities in AI-powered systems. This incident emphasizes the critical need for robust security measures as AI integration expands across platforms. Meanwhile, Sakana AI is focusing on recursive self-improvement, developing AI that can iteratively enhance itself, aiming to offer an alternative to the intense compute arms race among leading labs.