Someone downloaded an uncensored version of Qwen 3.8 27B locally and tested what happens when a powerful AI model has essentially no safety guardrails. The result? It reportedly answered a dangerous request that a properly aligned model would refuse. And that's where things get complicated. AI companies can build stronger safeguards into their models. Governments can create regulations. But once powerful open-weight models can be downloaded, modified, and run locally, how do you enforce those safeguards? You can't simply put the genie back in the bottle. As AI becomes more capable and increasingly accessible, the hardest question may no longer be: “How do we make AI safer?” It may be: “How do we maintain safety when anyone can remove the safeguards?” This is going to be one of the biggest AI debates of the next few years. 👀 Follow futurewalt.ai for more AI news and discussions. #ai #opensource #aisafety #technology