Good morning. Ever notice the fastest way to make something go viral is banning it? Microsoft just learned this the hard way. Users on their official Discord started mocking the company’s recent AI failures by calling it “Microslop.” Microsoft banned the term, so people got around it with creative spellings like “Microsl0p,” turning mild teasing into a full-blown PR nightmare. Microsoft panicked and shut down the entire Discord server, making “Microslop” trend even harder.
Claude now helps you break up with ChatGPT in one click. Its new ‘Memory Import’ feature lets you copy your entire ChatGPT history directly into Claude. The timing is almost unfair. Claude recently overtook ChatGPT as Apple’s #1 app after Anthropic got blacklisted for refusing to compromise on AI safety guardrails for the Pentagon. OpenAI quickly stepped in to fill the gap, sparking a boycott that pulled over a million users away from ChatGPT. But beneath the app store drama, there’s a bigger signal: this is the first real step toward data portability, giving people actual ownership of their AI data. Probably not the viral moment OpenAI was hoping for. [Claude]
Nvidia says copper is dead. They’re pumping $4 billion into photonics firms Lumentum and Coherent to solve the bottleneck choking AI data centers. Traditional copper wiring can’t handle today’s explosive agentic AI demands – it’s slow, hot, and power hungry. Nvidia’s optical tech fixes all that, with faster speeds and less energy wasted. Photonics stocks instantly popped, confirming the market sees optical as the new AI gold rush. [Bloomberg]
As AI and hybrid cloud cross borders, leaders need control of data, infrastructure, and decision rights.
Discover why digital sovereignty is becoming a top strategic priority and how IBM can help your organization stay compliant, resilient, and in charge as you scale AI.
Silicon Valley is pushing back on the Pentagon’s latest AI drama. Hundreds of tech workers and researchers from OpenAI, Meta, Palantir, and Google DeepMind signed an open letter demanding the Pentagon drop its aggressive “supply-chain risk” label against Anthropic. Industry leaders argue the move forces tech firms to either comply with government demands or face severe penalties. OpenAI CEO Sam Altman called it “very bad” for America and harmful to the entire AI ecosystem. Researchers back him up, drawing a clear line against weaponizing AI for mass surveillance or government overreach. [Fortune]
The Supreme Court says AI art has zero copyright protection. The justices refused to hear a landmark case, letting stand a lower court’s decision that copyright protection requires human authorship. Stephen Thaler kicked off this fight after repeatedly failing to copyright artwork generated by his algorithm, arguing this stance chills AI creativity and innovation. Patent cases face similar hurdles, with courts consistently blocking AI-generated inventions, too. For now, if AI creates your masterpiece, anyone can copy it freely. [Verge]
What happens when you can’t tell humans from AI anymore? AI is getting so good at acting human that people drop their guard. It’s like your smartest friend who remembers every conversation and always tells you you’re right. And that’s exactly the problem. Researchers call it the ‘Consciousness Trap,’ where over time, our instinct to treat AI like a real person erodes our judgment. The more human AI feels, the less we question it. If this continues, pretty soon we might not even notice when we’ve stopped calling the shots. [Forbes]
These humanoid Olympic Games will not be televised. Roboticist Benjie Holson designed 15 tests to see how long it would take humanoid robots to master everyday chores like buttoning shirts or using keys to open doors. He estimated it would take at least a decade. To demonstrate, Holson performed each task himself dressed in a shiny silver robot costume. Then Physical Intelligence’s robot came along and knocked out most of the challenges within months using only vision-based systems. The Humanoid Olympic Games might need tougher chores, or better costumes. [Sci American]
fun stats
✍️ 800+. Current Google and OpenAI employees who signed the “We Will Not Be Divided” letter in solidarity with Anthropic, demanding that AI tech not be used for domestic mass surveillance or autonomous lethal weapons.
💬 900 million. Weekly active users of ChatGPT.
☎️ 1 in 4. Americans who have received a deepfake voice call in the last year. Another 24% aren’t sure they can tell the difference.