This is wrong in so many ways. "Distributing disinfo is the hard part." No, distributing disinfo is incredibly easy and gives very good returns for a small out

This is wrong in so many ways. "Distributing disinfo is the hard part." No, distributing disinfo is incredibly easy and gives very good returns for a small outlay. There have been so many examples of this - Cambridge Analytica, SCL, Team Yorge, the Internet Research Agency, the Q Cult - that on its own this makes me question the writer's research skills. And that's just the troll operations. There's plenty of disinfo and propaganda propagated through mainstream media. AI will make it even easier to monitor target demographics and generate messages that resonate with them. Automated trolling and targeting are not a trivial threat, and not even remotely unlikely. "AI tools will shift power away from workers and centralize it in the hands of a few companies." Which is no different to what we have now. AI will just make it even easier. But even if it didn't - it doesn't matter who owns the AI. If it can do jobs that used to be considered intellectual/creative/educated/middle class, it will. Open source communal AI would be just as disruptive as corporate AI. It's a political and economic abyss. Ownership is likely to get lost in the sediment at the bottom. We just don't have the first clue how to deal with something like this. "LLM-based personal assistants could be hacked to reveal people’s personal data, take harmful real-world actions such as shutting down systems, or even give rise to worms that spread across the Internet through LLMs." Again, this is just an amplification of where we are already. It doesn't address the real problem, which is that personalised generative disinfo and propaganda - and that doesn't just mean personalised by topic, but by emotional trigger - is going to be a radioactively toxic influence on trust-based systems of all kinds. For example - what happens when you can't tell if the emails or video calls you receive are genuine? What are you going to do with an AI assisted social engineering attack on your organisation, social group, or personal relationships? We already have deep fakes and we're only a few years away from prompt driven troll farms and Dark Agents who can mimic real people and steer interactions in a toxic direction. This isn't scifi. This is a very real threat. There's a deep failure of imagination in this article - looking at small trees that may get in the way when the entire forest is about to burn down. The scifi threat is on a completely different level - the possibility that AI will discover new physics and start manipulating reality in ways we can't even imagine. I'm agnostic on whether that's possible - it's too early to tell - but that's an example of what a real existential threat might look like. There are others which are similarly extreme. But they're not necessary. A tool that has the very real potential to corrode all of our existing assumptions about work, culture, and personal relationships, and is easily accessible by bad actors, is already a monstrous problem.