The Daily AI Briefing - 11/04/2025

Welcome to The Daily AI Briefing, here are today's headlines! Today we're covering OpenAI's major memory upgrade for ChatGPT, Mira Murati's startup pursuing a record-breaking funding round, a clever technique to transform YouTube videos into SEO-optimized blogs, Microsoft's research showing AI still struggles with debugging, and a quick look at Google's latest Gemini model. Let's dive into the details. **OpenAI gives ChatGPT a memory boost** OpenAI has rolled out a significant update to ChatGPT's memory capabilities, enabling the AI assistant to automatically remember and reference information across all user conversations. Unlike previous versions where users had to specifically request information be remembered, the system now does this automatically, creating a more personalized experience. ChatGPT will now track users' preferences, interests, and dislikes across multiple conversations, then tailor its responses accordingly. This is particularly valuable for frequent users who previously had to switch between chats or repeatedly provide the same context. If you want to modify what ChatGPT knows about you, you can simply ask through a prompt. For privacy-conscious users, there's an option to opt out of the memory feature via settings or use temporary chat mode for conversations you don't want remembered. This update represents a significant step toward AI systems that genuinely get to know users over time, becoming increasingly useful through personalization. **Mira Murati's AI startup eyes record funding** Former OpenAI CTO Mira Murati's new venture, Thinking Machines Lab, is making waves in the investment world. The startup, led by several OpenAI veterans, is reportedly pursuing a massive $2 billion seed funding round at a valuation of at least $10 billion. This would make it one of the largest seed rounds in history, doubling what Murati initially targeted. What's particularly noteworthy is that Murati launched this startup just six months after leaving OpenAI, where she spent nearly seven years working on systems including ChatGPT. While specific details remain limited, the company aims to create "widely understood, customizable, and generally capable" AI systems. This funding pursuit joins another eye-popping AI investment – Ilya Sutskever's SSI is reportedly raising funds at a $30 billion valuation. Perhaps most striking is that both startups are attracting these massive investments without having public products or clear revenue roadmaps, highlighting investors' continued confidence in AI's future potential. **Turn YouTube videos into high-ranking blogs** A clever new workflow using Google's NotebookLM allows content creators to transform YouTube videos into SEO-optimized blog posts. The process begins by creating a notebook in NotebookLM and adding a YouTube video transcript as a source – either by direct link or pasted text. Next comes SEO preparation, identifying primary and secondary keywords relevant to your topic. For example, an AI automation piece might target phrases like "AI workflow tools" or "business process automation." You then craft a detailed prompt including these keywords and your desired content structure before generating the post. Final steps include enhancing the article with images, relevant links, and proper formatting before publishing. For technical topics, there's a pro tip: include a glossary section defining complex terms, which can improve search visibility for beginners looking for explanations of these concepts. This approach offers an efficient way to repurpose video content for written formats while maximizing search engine visibility. **Study: AI still struggles with software debugging** A new Microsoft Research study reveals a significant limitation in current AI systems: even the most advanced AI models struggle with routine software debugging tasks. Microsoft tested nine different LLMs, including Claude 3.7 Sonnet, by having them tackle 300 debugging issues from SWE-bench Li