I Measured 12 Korean Tokenizers. Two Were Borrowed.
Twelve tokenizers, one parallel corpus. The top three are recent Korean releases, while two others carry base vocabularies identical to Llama 3 and Mistral. Read more »
AI
Open Weight Licenses Aren’t Really Open
I measured the licenses and gating of the 100 most downloaded open models, separating what you can freely ship from what sits behind an access request. Read more »
The Top Spot in AI Model Downloads Isn’t a Chatbot
A direct tally of HuggingFace’s top 100 downloads puts a 2021 embedding model at number one, not a chatbot. Likes point the opposite way from downloads. Read more »
Why AI Agents Fail: Context Engineering 101
Context engineering decides what enters an AI model’s context window — the choice that makes the same model succeed or fail. It underpins reliable AI agents. Read more »
EU AI Content Labeling: The Aug 2 Deadline
EU AI Act Article 50 takes effect August 2, 2026, forcing labels on chatbots, deepfakes, and AI media — exposing where watermarking and C2PA break down. Read more »
World Models: AI’s Next Frontier, Explained
If LLMs predict the next word, world models predict a scene’s next state. Here’s how they work, and what Genie 3 and Marble actually change. Read more »
Prompt Injection: How AI Agents Get Hacked
Prompt injection exploits an LLM’s inability to tell trusted instructions from untrusted data: direct vs indirect attacks, the trifecta, and real defenses. Read more »
AI Data Center Power: The Grid Bottleneck
The real limit on AI data centers isn’t making power but connecting it — why two IEA reports disagree, and how 4 to 8 year grid queues set AI’s pace. Read more »
How Diffusion Models Generate Images and Video From Noise
A diffusion model creates images by adding noise, then reversing it—starting from pure static and denoising step by step until a clear picture appears. Read more »
AI Agent Benchmarks Are Breaking in 2026
Why soaring AI agent benchmark scores mislead — Berkeley’s exploit of 8 benchmarks, OpenAI’s SWE-bench retirement, and the pass@1 reliability gap. Read more »