Prompt Injection
@www.promptinjection.net
AI beyond the hype. Real insights, real breakthroughs, real methods. Philosophy, benchmarks, quantization, hacks—minus the marketing smoke. Injecting facts into https://www.promptinjection.net/
StatelessChatUI: One HTML file for direct LLM API access.
No installation, no server, no dependencies.
Double-click → runs. Open in browser → runs. Put on webserver → runs.
Connects directly to any OpenAI-compatible API.
Built for prompt engineering, debugging, teaching.
🧵
No installation, no server, no dependencies.
Double-click → runs. Open in browser → runs. Put on webserver → runs.
Connects directly to any OpenAI-compatible API.
Built for prompt engineering, debugging, teaching.
🧵
December 18, 2025 at 4:23 PM
StatelessChatUI: One HTML file for direct LLM API access.
No installation, no server, no dependencies.
Double-click → runs. Open in browser → runs. Put on webserver → runs.
Connects directly to any OpenAI-compatible API.
Built for prompt engineering, debugging, teaching.
🧵
No installation, no server, no dependencies.
Double-click → runs. Open in browser → runs. Put on webserver → runs.
Connects directly to any OpenAI-compatible API.
Built for prompt engineering, debugging, teaching.
🧵
When a Local LLM with Q2 Outperforms Cloud Q8: The Parameters Nobody Talks About
How AI Cloud Provider Defaults Can Sabotage Even The Best Models
www.promptinjection.net/p/ai-llm-whe...
How AI Cloud Provider Defaults Can Sabotage Even The Best Models
www.promptinjection.net/p/ai-llm-whe...
When a Local LLM with Q2 Outperforms Cloud Q8: The Parameters Nobody Talks About
How AI Cloud Provider Defaults Can Sabotage Even The Best Models
www.promptinjection.net
December 17, 2025 at 11:54 AM
When a Local LLM with Q2 Outperforms Cloud Q8: The Parameters Nobody Talks About
How AI Cloud Provider Defaults Can Sabotage Even The Best Models
www.promptinjection.net/p/ai-llm-whe...
How AI Cloud Provider Defaults Can Sabotage Even The Best Models
www.promptinjection.net/p/ai-llm-whe...
The 100-Trillion-Token X-Ray: What OpenRouter Reveals About Real AI Usage
Why the loudest debates about AI miss what's actually happening in production
www.promptinjection.net/p/the-100-tr...
Why the loudest debates about AI miss what's actually happening in production
www.promptinjection.net/p/the-100-tr...
The 100-Trillion-Token X-Ray: What OpenRouter Reveals About Real AI Usage
Why the loudest debates about AI miss what's actually happening in production
www.promptinjection.net
December 15, 2025 at 3:06 PM
The 100-Trillion-Token X-Ray: What OpenRouter Reveals About Real AI Usage
Why the loudest debates about AI miss what's actually happening in production
www.promptinjection.net/p/the-100-tr...
Why the loudest debates about AI miss what's actually happening in production
www.promptinjection.net/p/the-100-tr...
AI News Roundup: December 01 – December 12, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
AI News Roundup: December 01 – December 12, 2025
The most important news and trends
www.promptinjection.net
December 12, 2025 at 4:43 PM
AI News Roundup: December 01 – December 12, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The Quantization Cliff: When Does Compression Break Code Intelligence?
A Tetris coding experiment reveals where model compression crosses from viable to broken
www.promptinjection.net/p/ai-llm-the...
A Tetris coding experiment reveals where model compression crosses from viable to broken
www.promptinjection.net/p/ai-llm-the...
The Quantization Cliff: When Does Compression Break Code Intelligence?
Testing GLM 4.5 Air from 4-bit to 1-bit: A Tetris coding experiment reveals where model compression crosses from viable to broken
www.promptinjection.net
December 11, 2025 at 1:45 PM
The Quantization Cliff: When Does Compression Break Code Intelligence?
A Tetris coding experiment reveals where model compression crosses from viable to broken
www.promptinjection.net/p/ai-llm-the...
A Tetris coding experiment reveals where model compression crosses from viable to broken
www.promptinjection.net/p/ai-llm-the...
Llama 4 Scout: Total Disaster – or Misunderstood Workhorse?
From benchmark scandal to translation savant: the bizarre performance profile of Meta's most hated model
www.promptinjection.net/p/meta-llama...
From benchmark scandal to translation savant: the bizarre performance profile of Meta's most hated model
www.promptinjection.net/p/meta-llama...
Llama 4 Scout: Total Disaster – or Misunderstood Workhorse?
From benchmark scandal to translation savant: the bizarre performance profile of Meta's most hated model
www.promptinjection.net
December 8, 2025 at 3:02 PM
Llama 4 Scout: Total Disaster – or Misunderstood Workhorse?
From benchmark scandal to translation savant: the bizarre performance profile of Meta's most hated model
www.promptinjection.net/p/meta-llama...
From benchmark scandal to translation savant: the bizarre performance profile of Meta's most hated model
www.promptinjection.net/p/meta-llama...
Can Parameters Compensate for Aggressive Quantization?
When a 355B model with 1-bit dynamic quantization loses to a 106B model at 4-bit — and what it means for local LLM deployment.
www.promptinjection.net/p/can-parame...
When a 355B model with 1-bit dynamic quantization loses to a 106B model at 4-bit — and what it means for local LLM deployment.
www.promptinjection.net/p/can-parame...
Can Parameters Compensate for Aggressive Quantization?
When a 355B model with 1-bit dynamic quantization loses to a 106B model at 4-bit — and what it means for local LLM deployment. A Real-World Test.
www.promptinjection.net
December 6, 2025 at 2:47 PM
Can Parameters Compensate for Aggressive Quantization?
When a 355B model with 1-bit dynamic quantization loses to a 106B model at 4-bit — and what it means for local LLM deployment.
www.promptinjection.net/p/can-parame...
When a 355B model with 1-bit dynamic quantization loses to a 106B model at 4-bit — and what it means for local LLM deployment.
www.promptinjection.net/p/can-parame...
AI News Roundup: October 04 – November 30, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
AI News Roundup: October 04 – November 30, 2025
The most important news and trends
www.promptinjection.net
December 1, 2025 at 10:25 AM
AI News Roundup: October 04 – November 30, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The Safety Paradox: How RLHF Creates the AI Psychosis Problem It’s Meant to Prevent
www.promptinjection.net/p/ai-psychos...
www.promptinjection.net/p/ai-psychos...
The Safety Paradox: How RLHF Creates the AI Psychosis Problem It’s Meant to Prevent
When “Every Perspective Is Valid” Meets Vulnerable Minds
www.promptinjection.net
December 1, 2025 at 10:25 AM
The Safety Paradox: How RLHF Creates the AI Psychosis Problem It’s Meant to Prevent
www.promptinjection.net/p/ai-psychos...
www.promptinjection.net/p/ai-psychos...
How an AI-Secured Bank Was Robbed
A prompt injection demonstration in two acts
www.promptinjection.net/p/how-an-ai-...
A prompt injection demonstration in two acts
www.promptinjection.net/p/how-an-ai-...
How an AI-Secured Bank Was Robbed
A prompt injection demonstration in two acts
www.promptinjection.net
October 16, 2025 at 10:02 AM
How an AI-Secured Bank Was Robbed
A prompt injection demonstration in two acts
www.promptinjection.net/p/how-an-ai-...
A prompt injection demonstration in two acts
www.promptinjection.net/p/how-an-ai-...
AI Morality Without Safety Training? How Pretraining Bakes In Pseudo-Ethics
Why 'neutral' base models still moralize - and why two words can make it all collapse
www.promptinjection.net/p/ai-moralit...
Why 'neutral' base models still moralize - and why two words can make it all collapse
www.promptinjection.net/p/ai-moralit...
AI Morality Without Safety Training? How Pretraining Bakes In Pseudo-Ethics
From Crazy Carl to NSFW SVG's: Why 'neutral' base models still moralize - and why two words can make it all collapse
www.promptinjection.net
October 5, 2025 at 6:48 AM
AI Morality Without Safety Training? How Pretraining Bakes In Pseudo-Ethics
Why 'neutral' base models still moralize - and why two words can make it all collapse
www.promptinjection.net/p/ai-moralit...
Why 'neutral' base models still moralize - and why two words can make it all collapse
www.promptinjection.net/p/ai-moralit...
AI News Roundup: September 10 – October 03, 2025
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
AI News Roundup: September 10 – October 03, 2025
The most important news and trends
www.promptinjection.net
October 5, 2025 at 6:48 AM
AI News Roundup: September 10 – October 03, 2025
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
AI News Roundup: September 02 – September 09, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
AI News Roundup: September 02 – September 09, 2025
The most important news and trends
www.promptinjection.net
September 10, 2025 at 4:11 PM
AI News Roundup: September 02 – September 09, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
Remember when Blake Lemoine claimed LaMDA was sentient?
We recreated the entire dialogue structure and fine-tuned Qwen 4B - and now it responds like it has a soul.
We recreated the entire dialogue structure and fine-tuned Qwen 4B - and now it responds like it has a soul.
September 3, 2025 at 4:26 PM
Remember when Blake Lemoine claimed LaMDA was sentient?
We recreated the entire dialogue structure and fine-tuned Qwen 4B - and now it responds like it has a soul.
We recreated the entire dialogue structure and fine-tuned Qwen 4B - and now it responds like it has a soul.
AI News Roundup: August 26 – September 01, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
AI News Roundup: August 26 – September 01, 2025
The most important news and trends
www.promptinjection.net
September 2, 2025 at 6:52 PM
AI News Roundup: August 26 – September 01, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
GPT-5: Intelligence in the Fog
Brilliant via API – flat on ChatGPT. What you get is decided by an invisible router. And it often seems to be way off.
www.promptinjection.net/p/openai-gpt...
Brilliant via API – flat on ChatGPT. What you get is decided by an invisible router. And it often seems to be way off.
www.promptinjection.net/p/openai-gpt...
GPT-5: Intelligence in the Fog
Why GPT-5 on ChatGPT often isn’t the full GPT-5 - and why that’s more than just a technical problem
www.promptinjection.net
August 29, 2025 at 12:45 PM
GPT-5: Intelligence in the Fog
Brilliant via API – flat on ChatGPT. What you get is decided by an invisible router. And it often seems to be way off.
www.promptinjection.net/p/openai-gpt...
Brilliant via API – flat on ChatGPT. What you get is decided by an invisible router. And it often seems to be way off.
www.promptinjection.net/p/openai-gpt...
AI News Roundup: August 19 – August 25, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
AI News Roundup: August 19 – August 25, 2025
The most important news and trends
www.promptinjection.net
August 27, 2025 at 6:45 PM
AI News Roundup: August 19 – August 25, 2025
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
The most important news and trends
www.promptinjection.net/p/ai-llm-new...
Gemma 3 270M fine-tune on Windows with @UnslothAI.
It worked! 🥳
It worked! 🥳
August 24, 2025 at 2:35 PM
Gemma 3 270M fine-tune on Windows with @UnslothAI.
It worked! 🥳
It worked! 🥳
AI News Roundup: August 11 – August 18, 2025
www.promptinjection.net/p/ai-llm-new...
Topics:
AI robotics, billion-dollar funds, OpenAI drama, enterprise AI races, tiny efficient LLMs, breakthroughs in vision, Congress probes, no-code AI tools, cybersecurity shakeups, and the new Qwen image model.
www.promptinjection.net/p/ai-llm-new...
Topics:
AI robotics, billion-dollar funds, OpenAI drama, enterprise AI races, tiny efficient LLMs, breakthroughs in vision, Congress probes, no-code AI tools, cybersecurity shakeups, and the new Qwen image model.
AI News Roundup: August 11 – August 18, 2025
The most important news and trends
www.promptinjection.net
August 19, 2025 at 3:34 PM
AI News Roundup: August 11 – August 18, 2025
www.promptinjection.net/p/ai-llm-new...
Topics:
AI robotics, billion-dollar funds, OpenAI drama, enterprise AI races, tiny efficient LLMs, breakthroughs in vision, Congress probes, no-code AI tools, cybersecurity shakeups, and the new Qwen image model.
www.promptinjection.net/p/ai-llm-new...
Topics:
AI robotics, billion-dollar funds, OpenAI drama, enterprise AI races, tiny efficient LLMs, breakthroughs in vision, Congress probes, no-code AI tools, cybersecurity shakeups, and the new Qwen image model.
Imagen 4 from Google is now generally available.
Ran some first tests. Happy with the results. What do you think?
Can be used in Gemini API, Vertex AI, and Google AI Studio: aistudio.google.com
Ran some first tests. Happy with the results. What do you think?
Can be used in Gemini API, Vertex AI, and Google AI Studio: aistudio.google.com
August 15, 2025 at 5:52 AM
Imagen 4 from Google is now generally available.
Ran some first tests. Happy with the results. What do you think?
Can be used in Gemini API, Vertex AI, and Google AI Studio: aistudio.google.com
Ran some first tests. Happy with the results. What do you think?
Can be used in Gemini API, Vertex AI, and Google AI Studio: aistudio.google.com
AI News Roundup: July 28 – August 03, 2025
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
AI News Roundup: July 28 – August 03, 2025
The most important news and trends
www.promptinjection.net
August 3, 2025 at 3:01 PM
AI News Roundup: July 28 – August 03, 2025
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
Why Crawl4AI Might Be the Missing Link in Your LLM Stack
Open-Source Intelligence Crawling for Agents, RAG, and Beyond
www.promptinjection.net/p/why-crawl4...
Open-Source Intelligence Crawling for Agents, RAG, and Beyond
www.promptinjection.net/p/why-crawl4...
Why Crawl4AI Might Be the Missing Link in Your LLM Stack
Open-Source Intelligence Crawling for Agents, RAG, and Beyond
www.promptinjection.net
August 2, 2025 at 2:32 PM
Why Crawl4AI Might Be the Missing Link in Your LLM Stack
Open-Source Intelligence Crawling for Agents, RAG, and Beyond
www.promptinjection.net/p/why-crawl4...
Open-Source Intelligence Crawling for Agents, RAG, and Beyond
www.promptinjection.net/p/why-crawl4...
AI News Roundup: July 21 – July 27, 2025
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
AI News Roundup: July 21 – July 27, 2025
The most important news and trends
www.promptinjection.net
July 28, 2025 at 10:03 AM
AI News Roundup: July 21 – July 27, 2025
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
The most important news and trends
www.promptinjection.net/p/ai-news-ro...
When the Agent Butchers Your Prompt: Why AI Coding IDEs Often Deliver Worse Results than Direct Access
An Essay on Hidden Biases, Lost Control, and the Misunderstanding of the "Wrong Prompt"
www.promptinjection.net/p/when-the-a...
An Essay on Hidden Biases, Lost Control, and the Misunderstanding of the "Wrong Prompt"
www.promptinjection.net/p/when-the-a...
When the Agent Butchers Your Prompt: Why AI Coding IDEs Often Deliver Worse Results than Direct Access
An Essay on Hidden Biases, Lost Control, and the Misunderstanding of the "Wrong Prompt"
www.promptinjection.net
July 26, 2025 at 9:20 AM
When the Agent Butchers Your Prompt: Why AI Coding IDEs Often Deliver Worse Results than Direct Access
An Essay on Hidden Biases, Lost Control, and the Misunderstanding of the "Wrong Prompt"
www.promptinjection.net/p/when-the-a...
An Essay on Hidden Biases, Lost Control, and the Misunderstanding of the "Wrong Prompt"
www.promptinjection.net/p/when-the-a...
Run Local AI with UI in Minutes – No Setup, No Coding
A minimal guide to running offline AI with zero setup using Llamafile on Windows.
www.promptinjection.net/p/run-local-...
A minimal guide to running offline AI with zero setup using Llamafile on Windows.
www.promptinjection.net/p/run-local-...
Run Local AI with UI in Minutes – No Setup, No Coding
A minimal guide to running offline AI with zero setup using Llamafile on Windows.
www.promptinjection.net
July 24, 2025 at 11:47 AM
Run Local AI with UI in Minutes – No Setup, No Coding
A minimal guide to running offline AI with zero setup using Llamafile on Windows.
www.promptinjection.net/p/run-local-...
A minimal guide to running offline AI with zero setup using Llamafile on Windows.
www.promptinjection.net/p/run-local-...