

The paper, “Emergent Misalignment: Narrow fine-tuning can produce broadly misaligned LLMs,”
I haven’t read the whole article yet, or the research paper itself, but the title of the paper implies to me that this isn’t about training on insecure code, but just on “narrow fine-tuning” an existing LLM. Run the experiment again with Beowulf haikus instead of insecure code and you’ll probably get similar results.
Exactly. Trump. Not the states that voted against him. If you want to retaliate, retaliate against Trump. Make him, and the people that voted for him feel the pain.
Look at the things he’s doing domestically. There’s nothing he and his MAGA dipshits love more than sticking it to the blue states. So, Canada doing something that predominantly hurts blue states isn’t going to hurt Trump at all. Quite the contrary, it’s going to give him huge hard-on.
So, if your goal is to help Trump, well then carry on.