📝 TL;DR Starting with Firefox 148, Mozilla is adding a dedicated AI Controls panel that lets you turn off all current and future AI features in one click, or pick and choose the ones you want. This is the first mainstream browser that makes AI truly optional, not something quietly forced into your workflow. 🧠 Overview AI is creeping into every browser and app, often in ways users do not really notice until things feel noisy or invasive. Mozilla has heard the pushback and is taking a different route, shipping a central AI Controls section in Firefox that includes a master “Block AI enhancements” switch plus granular toggles for each AI feature. It is a clear positioning play, yes, Firefox is becoming a “modern AI browser,” but only for people who actually want that. Everyone else gets a clean, AI free experience by default if they choose. 📜 The Announcement From Firefox 148 on February 24, desktop users will see a new AI Controls section in Settings. From that single screen, you can: - Hit one switch to block all AI “enhancements” now and in the future - Or selectively turn off things like the sidebar chatbot, AI translations, AI tab grouping, alt text generation for PDFs, and AI powered link previews Mozilla is framing this as a response to user feedback, especially from people who said they wanted “nothing to do with AI” in their browser, alongside others who do want useful AI but on their own terms. ⚙️ How It Works • Central AI Controls page - Firefox 148 adds a dedicated AI Controls section in Settings so you can see every AI feature in one place instead of hunting through multiple menus. • Master “Block AI enhancements” toggle - Turn this on to disable all existing AI features and prevent new ones from being surfaced or suggested in future releases. • Per feature toggles - If you like some AI but not all, you can turn off specific tools such as the sidebar chatbot, AI based translations, tab grouping suggestions, link previews, or PDF alt text.

📝 TL;DR A former Google engineer has been found guilty of stealing thousands of confidential AI supercomputing files to benefit companies tied to China. This is one of the first big criminal cases where the core prize was AI infrastructure, not just regular software. 🧠 Overview A US federal jury in San Francisco convicted former Google software engineer Linwei “Leon” Ding on 14 counts of economic espionage and theft of trade secrets. Prosecutors said he secretly copied thousands of pages of internal documents covering Google’s AI supercomputing stack, then used them while working with and pitching China based AI startups. The case is being held up by US officials as a warning shot in a new era of AI focused economic espionage, especially around chips and infrastructure that power large models. 📜 The Announcement According to the Justice Department, Ding joined Google in 2019 and worked on the company’s cutting edge AI infrastructure team. Starting in 2022, he began quietly uploading confidential design documents from Google’s internal network to a personal cloud account, including details on Tensor Processing Units, GPU clusters, and SmartNIC networking that power large scale AI training and inference. At the same time, he was secretly on the payroll of a China based AI startup and later founded his own AI company overseas, telling investors he could replicate Google’s supercomputing setup. After an 11 day trial, the jury convicted him on seven counts of economic espionage and seven counts of trade secret theft. He now faces the possibility of decades in prison and multi million dollar fines at sentencing. ⚙️ How It Works • What he stole - Thousands of files describing Google’s AI data center guts, including chip configurations, networking topologies, and software that ties it all together into an AI supercomputer. • How he stole it - He allegedly used his internal access to quietly upload documents from secure Google systems to a personal cloud account over many months, blending in with normal work activity.

Our new Weekly Roundup Newsletter is out now! You can find it here inside the classroom. Would love to hear your thoughts!

It’s the end of January. Did you actually become a different person this month…or did you just think about changing? No judgment. No shame. Just data. Because results don’t come from motivation. They come from identity shifts. So ask yourself: What did I do this month that the old me wouldn’t have done? What standard did I raise? What excuse did I stop tolerating? If nothing changed, that’s okay. But don’t lie to yourself about it. February belongs to whoever decides differently. Where did you win? Where did you stall? And if you are feeling brave drop your answers below so we can cheer you on and hold you accountable!

Most people don’t feel confused because they lack information. They feel confused because they have too much of it, arriving too quickly, from too many directions, without structure. Tabs pile up. Articles get skimmed. AI gives fast answers that sound confident, but don’t quite settle the question. So you move on… while still feeling uncertain. That’s the real problem this post is about. ---------- THE REAL ISSUE ---------- We’re trying to solve research problems with answer-style tools. Most AI interactions are optimized for speed. You ask a question, you get a response, and you move on. That works well for simple tasks, but it breaks down the moment something gets complex, uncertain, or high-stakes. When topics require context, comparison, trade-offs, or judgment, quick answers don’t reduce confusion. They often create false clarity. Things sound resolved, but the understanding underneath is thin. That’s why people still feel unsure even after “getting an answer”. ---------- WHY OVERWHELM KEEPS GROWING ---------- Information overload isn’t caused by too much content. It’s caused by a lack of synthesis. We collect pieces of information, but rarely step back to see how they fit together. We open more tabs, not because we want more data, but because we’re trying to find certainty. Instead, we end up with fragments and no coherent picture. This is mentally exhausting. Not because thinking is hard, but because unstructured thinking is. Your brain keeps the question open in the background, pulling attention even when you’re doing something else. Over time, this leads to decision fatigue, delay, or avoidance. ---------- THE SHIFT THAT NEEDS TO HAPPEN ---------- The key shift is simple, but subtle: Moving from asking AI for answers to using AI to support structured understanding. That doesn’t mean more prompts. It means a different mental model. Instead of “tell me the answer,” the question becomes “help me explore this properly”. Research isn’t a response. It’s a process. It involves gathering sources, comparing perspectives, identifying patterns, and understanding trade-offs. When that process is missing, confidence is fragile.