Butch Coolidge @vulnerablecodes

Anthropic just pulled the biggest bait-and-switch in AI history. The "privacy-first" ChatGPT alternative just forced users to choose: Accept our new terms or lose access forever. The terms? Complete betrayal of their founding values. Here's why they did it: 🧵

Today, we are covering the 4 stages of building LLMs from scratch to make them applicable for real-world use cases. We'll cover: - Pre-training - Instruction fine-tuning - Preference fine-tuning - Reasoning fine-tuning The visual summarizes these techniques. Let's dive in!

In 1983, Paul McCartney gave Michael Jackson business advice that would make MJ billions... And destroy their friendship forever. The story behind this $47.5 million power move reveals the most ruthless lesson in music business history: 🧵

Ollama v0.11.8 now enables flash attention by default on OpenAI's gpt-oss 20B and 120B models for systems that support it!

Hands-On Large Language Models. Code examples covering everything from language model basics to fine-tuning in ready-to-run notebooks. 100% Opensource.

Anthropic spent $100M+ on getting Claude to 75%~ success rate on SWE-Bench (far from saturated)... Bittensor subnet 62 (@ridges_ai) has permissionlessly incentivized thousands of miners globally to achieve 66.8% on SWE-Bench from zero in 45~ days. Exponentially fast. $TAO.…

A 21-year-old just built an AI system that outperforms Anthropic's $100M+ investment. @ridges_ai: 72% on SWE-Bench in 46 days for $650K @AnthropicAI: 75% after hundreds of millions and years of development Meanwhile VCs keep writing nine-figure checks to the same old faces.…

Nano banana + Linah AI + n8n = Ad Factory This system pumps out TikTok/FB/Insta video ads on autopilot using the latest AI video models. - No actors. - No editors. - No overpriced agencies. Just endless, scroll-stopping UGC-style ads at scale. Perfect for e-com brands &…

A new open-source training method makes AI 94.8% secure on code generation benchmarks beating Anthropic's Claude Sonnet 4 and OpenAI's o4-mini. The "PurpCode" paper from @UofIllinois won the Amazon Nova AI Challenge. The model they built is the proof, but the open-source…

🚨🛰 Project Kuiper update! Another batch of production satellites for our low Earth orbit internet constellation is on its way to Cape Canaveral, FL. These satellites, built to withstand the harsh conditions of space and the journey there, will be processed upon arrival to get…

keyboard_arrow_left Previous keyboard_arrow_right Next

5 levels of evolution of AI Agents. Over the last few years, we’ve gone from simple LLMs → to fully-fledged Agentic systems with reasoning, memory, and tool use. Here’s a step-by-step breakdown. 1) Small context window LLMs - Input: Text → LLM → Output: Text - Early…

10,000+ people joined the OpenAI to Z Challenge to explore how AI can push the archaeological frontier in the Amazon. 🌱 The winner is Team Black Bean, which used deep learning on public LiDAR + satellite data to build maps that surface what's under rainforest canopies.

VirusTotal shuts down. What are you using to triage and analyze suspicious files or URLs?

It’s rare for competitors to collaborate. Yet that’s exactly what OpenAI and @AnthropicAI just did—by testing each other’s models with our respective internal safety and alignment evaluations. Today, we’re publishing the results. Frontier AI companies will inevitably compete on…

I’ve been working on something new: 📚 Build a Reasoning Model (From Scratch). The first chapters just went live! (The book will cover topics from inference-time scaling to reinforcement learning)

Our new Threat Intelligence report details how we’ve identified and disrupted sophisticated attempts to use Claude for cybercrime. We describe a fraudulent employment scheme from North Korea, the sale of AI-created ransomware by someone with only basic coding skills, and more.

Urgent! Just by receiving an image in your iPhone or Mac, your device can be FULLY compromised. Update your devices ASAP! Also recommended: • Disable auto download of images in Telegram; • Disable the auto download of images in WhatsApp; • Disable iMessage for complete or,…

China's new robot mall looks like something straight out of Blade Runner. Humanoid robots greet customers. Androids pour coffee. AI assistants negotiate prices. You can take one home today for $278 - less than your iPhone. But here's what they're not telling you about these…

this morning @RayFernando1337 woke me up at 8am to do a podcast.. he caught me SUPER caffenated and i just went off... recording just dropped! Marketing 101 for nerdy developers with sweet app/saas ideas but no marketing skills youtube.com/watch?v=ECsOBs…

The most dangerous man in AI isn't Elon or Altman It's the 28-year-old who built the training system for OpenAI, Google and U.S. Department of Defense. At 19, he dropped out of MIT to solve the dirtiest job in AI. Here's how Alexandr Wang outplayed the entire AI industry: 🧵