If you have a somewhat decent Graphics card, you can run AI applications locally... Image and video generation, AI chatbots that are stunningly capable- if I wanted to I could generate all sorts of fake video and audio of famous people saying and doing things... I could have my own local fake news factory.

I had a computer built specifically for handling 3D rendering, large datasets, and AI workloads- the requirements for the 3 categories overlap... I was recently testing out local chatbots- specifically QWEN3.6... its like a somewhat dumbed-down Claude, but remarkably capable. These local LLM run entirely on your own machine, completely private. They still have guardrails- the original releases have self censoring- QWEN is from China and will censor itself if you ask it how to make explosives, or about the Tiananmen square massacre...

People of course quickly figured out ways to remove all guardrails, the trick is removing the inhibitions without dumbing it down- lobotomizing it - its called 'Abliteration' these uncensored motels are freely available- I downloaded an Abliterated QWEN3.6 and tested it out...

I asked it how to make a certain explosive I learned how to make as a kid- its one that was used a lot by terrorists because its so easy to make (its also easy to accidentally detonate as many terrorists learned the hard way, by experiencing rapid disassembly of their body). It requires just 3 ingredients that are freely available at hardware stores and beauty salons... The uncensored LLM had no qualms explaining to me how to make the explosive... it's instructions were correct, and even gave me tips I didn't know about.

I have little doubt that the LLM would just as happily explain to me how to make Ricin or other weapons of mass destruction- even though its 100% private, I didn't feel comfortable asking to test it... I am sure it would give me great pointers on how to genetically modify a virus to manufacture a pandemic... the level of detail only limited by its training...

We are in a world where, as terrifying as it is, a motivated grad student with 100's of thousands of dollars to spare could build a pandemic... and with LLMs the cost of entry to making terrible weapons has been dramatically lowered- no longer do you need special training or education.

It is just a matter of time before we see horrific terrorist acts enabled by AI assistance... I am shocked its not happening already... and the scale of the danger will only grow super-exponentially...

I am not sure how this will turn out, but it certainly wont be the techno-utopia some people are trying to sell us...