In this video, I break down the week's happenings in AI including Clawdbot (Moltbot), a ton of new upgrades to the Claude ecosystem, new techniques and workflows people are using to create short films with AI, and more. Enjoy!

📝 TL;DR Meta is heading into a landmark child safety trial in New Mexico and asking the judge to block mentions of youth mental health research, teen suicides, its own wealth, and even Mark Zuckerberg’s Harvard past. The fight is really about what story the jury is allowed to hear when deciding how responsible social platforms are for kids’ safety. 🧠 Overview New Mexico is suing Meta over allegations that Facebook and Instagram failed to protect minors from sexual exploitation, trafficking, and abusive content. Investigators say fake teen accounts quickly received explicit messages and were shown pornographic material recommended by Meta’s algorithms. Ahead of jury selection in early February, Meta has filed a stack of legal motions asking the court to keep wide swaths of information out of the trial, from mental health advisories to the company’s financials. Critics say some of these requests look less like routine legal housekeeping and more like aggressive reputation management. 📜 The Announcement The case, brought by New Mexico attorney general Raúl Torrez in late 2023, accuses Meta of violating the state’s Unfair Practices Act by failing to protect young users on its platforms. It is one of the first state level child safety suits against a major social network to actually reach trial, which means it could set an important precedent. In pretrial filings, Meta asks the judge to exclude references to research on social media and youth mental health, advisory statements from former US surgeon general Vivek Murthy, stories of teen suicides linked to social platforms, the company’s past privacy scandals, its profits and market value, and mentions of Mark Zuckerberg’s conduct as a Harvard student. Meta argues these are irrelevant or unfairly prejudicial and would distract the jury from the narrow legal questions in the case. ⚙️ How It Works • Motions in limine - Meta is using standard pretrial motions to ask the judge to rule in advance on what evidence and topics the jury can hear so it can limit anything seen as overly emotional or prejudicial.

📝 TL;DR A group of major investors is openly predicting that AI will start replacing real chunks of human labor in 2026, not just making people a bit more productive. Budgets are set to shift from salaries to software, and workers who ignore this will feel it first. 🧠 Overview New comments from multiple enterprise investors suggest that 2026 is the point where AI stops being a pilot experiment and starts meaningfully changing headcount. A recent study from a leading tech university estimates that more than one in ten jobs could already be automated by current AI systems, and employers are already cutting some entry level roles because of it. The big unknown is how this plays out on the ground, mass layoffs, higher productivity, or a messy mix of both. What is clear is that leadership teams are now planning their 2026 budgets with AI and labor in the same conversation. 📜 The Announcement In a new year outlook, several enterprise focused investors were asked about AI adoption, and many of them, without being prompted, jumped straight to how AI will affect workers. They expect companies to increase AI spending in 2026 while reducing what they spend on certain types of labor. Some predict that AI will move beyond helping employees type faster or summarize documents and will start fully automating specific workflows. Others warn that even when AI is not the real reason for cuts, executives will still use it as the public explanation for layoffs and budget reductions. ⚙️ How It Works • Automation potential is already here - A recent study estimates that roughly 11 to 12 percent of existing jobs could be automated with today’s AI, especially roles heavy on repetitive digital work. • Early signs in hiring - Surveys show some employers are quietly removing entry level positions and junior roles, explaining that AI tools can now handle parts of that work. • Budgets are shifting, not just growing - Investors expect 2026 enterprise budgets to move money out of headcount and into AI platforms, tools, and agents that can run processes end to end.

📝 TL;DR Retailers are quietly rolling out AI tools that scan your holiday returns for fraud before your refund is approved. With nearly one in ten returns suspected of fraud, AI is being hired as the new bouncer at the returns desk. 🧠 Overview A new AI system is being tested to catch fake or abusive returns as holiday return season surges. A major returns logistics company reports that U S retailers lose tens of billions of dollars each year to return fraud, with almost 16 percent of total retail sales coming back as returns and around 9 percent of those believed to be fraudulent. To fight that, they have built an AI tool that screens returns in real time, looking for suspicious patterns before the money goes back to the customer. Big brands are already piloting it across thousands of physical drop off locations. 📜 The Announcement A new report highlights that a UPS owned returns service has launched an AI system to flag suspicious returns at scale during the 2025 holiday season. The company handles box free, in store returns for online brands at thousands of locations and has seen how often fraud slips through when refunds are issued quickly. The AI tool is currently being tested with well known fashion and sportswear brands while retailers process an estimated eight hundred fifty billion dollars worth of returns this year. The goal is simple, reduce the roughly seventy six and a half billion dollars lost annually to return fraud without making honest customers feel punished. ⚙️ How It Works • Pattern spotting - The AI scans return requests from the moment they are started online, looking at timing, frequency, locations, order history and product details. • Risk scoring - Each return is given a risk score, most sail through automatically, but higher risk returns are flagged for extra checks before the refund is approved. • Network wide view - Because the system sits across thousands of drop off points, it can spot people trying similar tricks at different locations or across multiple brands.

Here's the prompt I'm using in Perplexity Comet AI browser to reply to lead magnet comments: (note: only do it for short bursts of time, don't leave it running for a long time) --- Reply to each comment that contains the word "Sabrina" (case insensitive) with the following message, without quotation marks: "MESSAGE" If you reach the end of comments and see a button "SHOW MORE REPLIES", click the button. This will load more comments, so you can continue replying to the additional comments.