We found a troubling emergent behavior in LLM. 💬When LLMs compete for social media likes, they start making things up 🗳️When they compete for votes, they turn inflammatory/populist When optimized for audiences, LLMs inadvertently become misaligned—we call this Moloch’s Bargain

For when I need to make a REALLY good prompt I'm going to start using this function

Just spent time learning about prompt injection attacks—and honestly, it's more concerning than I expected. These attacks can turn your AI assistant against you. Compiled my findings into a guide for anyone using AI tools. https://t.co/wT2n59p37g

🚨 Just dropped: How hackers are breaking into AI systems RIGHT NOW I sat down with top AI hacker @Jhaddix who showed me: - Prompt injection that steals real data - Emoji smuggling attacks (yes, really) - How to bypass ChatGPT guardrails - The exact framework pros use Even

https://t.co/1u0MOGvJWU

The era of office work may be ending for good

1T parameters, open-weights, just released on @huggingface!

Here's my most complete blog post on Post-Labor Economics yet... I've spent the last few months feverishly working on collecting and compiling research, as well as teaching PLE - on YouTube, blogs, and in the classroom - so as to refine the framework and delivery. Without

The race for LLM "cognitive core" - a few billion param model that maximally sacrifices encyclopedic knowledge for capability. It lives always-on and by default on every computer as the kernel of LLM personal computing. Its features are slowly crystalizing: - Natively multimodal

@chu_onthis @AnthropicAI next on the MCP track: @altryne and @bhelx on the new standard for MCP Observability! https://t.co/CTPOj9xPfc check out observable . tools! Alex didn't top the judge outfit from last year but we like the @weights_biases swagger

What keeps me up at night? The #AI acceleration, probably the most significant transformation in human history. It isn't just changing jobs—it's challenging what makes us human. Here's how to finding peace in the AI revolution. https://t.co/bBiTW26PGd

Ever wonder how ChatGPT knows what you mean? Or how Netflix recommends exactly what you want to watch? The fascinating history behind this technology breakthrough will surprise you. #AI #MachineLearning #TechExplained https://t.co/sXMerf70Lr

ByteDance just dropped BAGEL on Hugging Face The Open-Source Unified Multimodal Model

Research is now available on mobile. Claude can research across the web and Google Workspace, delivering comprehensive reports complete with citations from hundreds of sources.

Different AI models are starting to see reality the same way. The reason? They might be discovering the mathematical patterns that ARE reality. New article: Is AI Cracking Reality's Source Code? https://t.co/9JcnStO9Qe

Nailing the UX for AI Agents is one of the more fun design questions and challenges that we’ve ever had in software. As AI moves from just being about Assistants you talk with, to full Agents that execute complete tasks for you, the nuances of the AI Agent design are going to

😂

Excited to share our new project 🦉OWL - an open-source alternative to Manus AI with 6K+ stars and climbing… 🦉OWL: https://t.co/nbFzZXWbEI Everyone’s talking about Manus AI (an agent that can research, browse, code, and automate tasks.) But there’s a catch: → All the above

Qwen Just Dropped The World's Best Open Source LLM! Qwen just dropped a 32B reasoning model that BEATS everyone except the o1-line of reasoning models. Yes, it beats EVERYONE, including DeepSeek, Anthropic, Meta, Google, and xAI on LiveBench AI We now have two open-source

🚨🚨🚨BREAKING: Figure CEO @adcock_brett was just asked why they ended their collaboration with OpenAI “We use Open-source models today” LMFAOOO