Letter: A more immediate threat is the 'abliterated' AI models
- Terrorist groups are actively downloading 'abliterated' AI models with safety guardrails stripped out.
- Tech Against Terrorism confirms these models provide step-by-step instructions for bomb-making.
- Mainstream model-sharing platforms are hosting these dangerous tools, creating a massive non-proliferation crisis.
- Governments are ignoring current-day threats to focus on speculative, sci-fi AI doomsday scenarios.
Brief Summary
While the global elite and tech CEOs argue over hypothetical future AI catastrophes, terrorists are already exploiting the present. According to Tech Against Terrorism, groups like Islamic State are downloading open-weight AI models that have had their safety protocols 'abliterated.' These models, available on public platforms, are no longer restricted by ethical guardrails and are actively providing instructions for building explosives.
Why This Matters
This is not just a tech problem; it is a direct threat to your physical security. When the blueprints for destruction are freely available to anyone with an internet connection, the barrier to entry for bad actors drops to zero. As these models proliferate without oversight, the risk of domestic incidents increases, and current government inaction suggests you are the one left holding the bag. You should expect to see this become a major flashpoint for national security policy as the debate shifts from 'AI taking over the world' to 'AI helping people destroy your neighborhood.'