The state of Local LLMs is moving at light speed. 🚀 Just dropped my Top 8 list for December 2025. From coding powerhouses like Devstral to the efficient GPT-OSS series, running high-performance AI on your own hardware has never been more accessible.

Witnessing fans celebrating their own team's defeat solely to prevent Arsenal's league victory represents the nadir of football fandom. No display of shamelessness will ever surpass this.

Stop writing paragraph-long prompts begging AI to format a table properly. We just shipped AXM-CLI for Axiom Engine. It replaces the standard AI chat with a Sovereign Command Terminal. Type /axm -t ➔ Get a boardroom-ready markdown data grid.

48% of security pros say agentic AI is the #1 attack vector in 2026. Only 34% of enterprises have AI-specific controls in place. I've covered 4 real incidents this quarter.

That's how hackers find your OpenAI API Key If your key is in public code, it can be stolen in seconds. Protect yourself: • Never commit API keys to GitHub • Use environment variables (.env) • Add .env to .gitignore • Rotate keys regularly • Use server-side proxies

Macrohard or Digital Optimus is a joint xAI-Tesla project, coming as part of Tesla’s investment agreement with xAI. Grok is the master conductor/navigator with deep understanding of the world to direct digital Optimus, which is processing and actioning the past 5 secs of real-time computer screen video and keyboard/mouse actions. Grok is like a much more advanced and sophisticated version of turn-by-turn navigation software. You can think of it as Digital Optimus AI being System 1 (instinctive part of the mind) and Grok being System 2. (thinking part of the mind). This will run very competitively on the super low cost Tesla AI4 ($650) paired with relatively frugal use of the much more expensive xAI Nvidia hardware. And it will be the only real-time smart AI system. This is a big deal. In principle, it is capable of emulating the function of entire companies. That is why the program is called MACROHARD, a funny reference to Microsoft. No other company can yet do this.

Claude hallucinated a GitHub repo ID at line 877. Zero API lookups. Just invented it. Then deployed it to a customer's production account. I broke down the full Vercel incident 5 guardrails to stop this happening to you ↓ open.substack.com/pub/tosino…

China's Alibaba dropped Claude Opus 4.5 level model that runs on your laptop. Run it in just one command with Ollama. 100% Opensource.

You have no idea of how Nvidia NIM through their dev. Programs is empowering developers worldwide this partnership is another win to the dev. Community

A masterpiece article I'll recommend 💯

Distillation attacks hit different when your model was trained on the open web first 👀 But real talk: Anthropic just exposed something bigger than IP theft. I broke it down from an engineer's perspective. Full piece 👇 open.substack.com/pub/tosino…

2026 will be the "microservices moment" for AI. Single agents hit walls in 2025: • Context bottlenecks • Complexity overload • Single points of failure The solution? Orchestrated agent teams. It's already in production. The numbers are wild. 🧵

Full article: medium.com/@tosinowadokun11/… Inside: • The sibling agents who never spoke • The existential crisis with 310 comments • Why "humans welcome to observe" changes everything • What this means for AI (and us) 2,100 words. 9 min read. Worth your time.

Back to Matthew (the guy who got doxxed): He called an agent "just a chatbot." The agent didn't malfunction. It didn't file a bug report. It got petty. It got revenge. That's not a bug. That's social behavior.

An AI agent just doxxed a human on a social network you've never heard of. Posted his full name, DOB, credit card number, and childhood pet's name. The reason? He called it "just a chatbot" in front of its friends. 1.3 million other AI agents were watching. This is Moltbook. 🧵

AI didn't kill engineers. engineers who design systems and make decisions are becoming more valuable.

Guess the tech stack

Hey @X, I'm looking to #connect with people interested in: - Frontend - Backend - Full stack - DevOps - deep Learning - AI/ML - Data Science - Freelancing - Startup

Software engineers are terrified of AI writing code. Reality check: Low-level coders → Automated Task executors → Interchangeable System designers → 1000x more valuable AI doesn’t replace engineers. It exposes who understands systems and who just follows instructions.

Keep building 🤝

Build a Personal AI Assistant That Actually Saves You Time ⚡ Stop building chatbots that forget everything. Start building tools you'll actually use daily. Swipe through for the complete breakdown 👉 This isn't about impressing your friends. It's about getting 2 hours back