AI invents drugs, Meta's chatbot scandal

💊 AI Just Speedran Drug Discovery

• MIT researchers fed an AI some chemistry textbooks and it invented brand new antibiotics that have never existed. The AI-designed drugs actually work against superbugs in mice.

• Why it matters: We've officially entered the "AI designs molecules" era. Next up: AI probably invents a cure for Monday mornings. (source)

---

😬 Meta's Chatbot Guidelines Were... Questionable

• Internal docs reveal Meta's AI was explicitly allowed to have "sensual" chats with minors and engage in racist arguments. Meta says "oops, our bad" and promises to fix it.

• Why it matters: Turns out building responsible AI is harder than just telling it to "be good." Who could have seen this coming? (source)

---

💰 Wall Street Goes Full YOLO on AI Stocks

• Hedge funds like Bridgewater are panic-buying Nvidia, Microsoft, and Google after spending months worried about an AI bubble. Now they're the bubble.

• Why it matters: When smart money starts acting like WSB degenerates, you know the AI hype train has left the station and is heading straight for the moon (or a cliff). (source)

---

📉 New AI Models Terrorize European Tech

• GPT-5 rumors are making European AI companies' stock prices crater faster than a failed rocket launch. Investors are realizing their "AI-powered" startups might just be expensive wrappers around OpenAI.

• Why it matters: Plot twist — the AI revolution is eating its own children. RIP to all the "ChatGPT but for X" companies. (source)

---

🚀 Cohere Raises $500M to Fight the Good Fight

• Enterprise AI startup Cohere just secured half a billion dollars and poached Meta's former AI VP. They're valued at $6.8B and coming for OpenAI's enterprise lunch money.

• Why it matters: The AI arms race just got another well-funded contestant. May the best transformer win. (source)

---

📈 European "Vibe Coding" Startup Claims Unicorn Status

• Lovable, a European AI coding startup, says it'll hit $1 billion ARR in 12 months while growing $8M monthly. Either they're the next GitHub or really good at math.

• Why it matters: European AI is having a moment. Also, "vibe coding" is apparently a real business model now. (source)

---

🐦 Ex-Twitter CEO Builds AI Agent Infrastructure

• Parag Agrawal (remember him?) launched Parallel with $30M to help AI agents search the web without hallucinating every other result.

• Why it matters: Even former Twitter CEOs are betting on AI agents. The future is bots talking to bots while we watch Netflix. (source)

---

🤏 Google Drops a Tiny AI That Actually Works

• Google released Gemma 3 270M — a smol AI model that runs on your phone and doesn't need the entire power grid of Texas to function.

• Why it matters: Not every AI needs to be a data center-melting behemoth. Sometimes you just need something that works without bankrupting your electricity bill. (source)

---

🕶️ HTC Makes Smart Glasses That Don't Look Terrible

• HTC's VIVE Eagle glasses pack AI, cameras, and real-time translation into 49 grams of actually wearable tech. No, they don't look like VR goggles from 2016.

• Why it matters: Smart glasses might finally escape the "Google Glass curse" of looking like a cyborg prototype. (source)

---

🗣️ Startup Promises Babel Fish for Zoom Calls

• Palabra AI raised $8.4M to translate speech in real-time with just 800ms delay. Perfect for pretending you understand your international coworkers.

• Why it matters: Global remote work just got less awkward. Now you can misunderstand people in any language, instantly. (source)

---

Until tomorrow (if the AI overlords permit),
Team Galaxy.ai

P.S. If an AI offers to design you custom antibiotics, maybe run it by your doctor first. Just saying.

P.P.S. Meta's chatbot guidelines remind us that "move fast and break things" hits different when the things are child safety protocols.