In the latest AI-Podden news episode, our hosts Ather Gattami and Anders Arpteg, discussed insights from a recent Stockholm AI event with Mistral and Anthropic, covering debates on open-source vs. open weights, safety-first agent design, and the rise of "context engineering." They reviewed major lawsuits involving Google, OpenAI, and Anthropic, highlighted Google's new Nano Banana image generator, and reflected on AGI's limits, arguing that today's AI excels at knowledge management but still lacks true reasoning and autonomy, with future progress hinging on new architectures, massive compute, and energy considerations.