Cindʎ Xiao 🍉

RE: https://infosec.exchange/@mttaggart/116371937587992264

I think the way I would put it is:

1) The point of the AI project is ideological; the goal is to reshape industries such that we are dependent on AI companies' products, and to destroy free and open knowledge such that we are dependent on these products for thought and reasoning. We see an injection of AI into cybersecurity, while simultaneously drawing money and resources away from (boring) efforts that would actually broadly improve cybersecurity.¹ We see an injection of AI into knowledge acquisition, while simultaneously polluting the landscape of the internet as a useful source of knowledge. Both are in service of the same ideological project, and working towards the same goal.

2) The touted usefulness of AI for programming and cybersecurity is directly funding the project to expand it everywhere else, where it is causing massive harms to civil society, individuals' mental health, and the information landscape. You or your company paying for these products is keeping investment money flowing and extending the runway, for AI companies to reach that point of "indispensability". There is no divorcing your cool shiny toy from the creation of AI deepfakes that destroy democracy, or the AI psychosis that destroys lives. This is because the AI companies are pursuing an ideological project that ultimately has nothing to do with improving people's work or their lives; the leaders of these companies have loudly and publicly said that very clearly. You are laundering the reputation of these companies and keeping them alive, when the only moral option is to destroy them.

I've said this elsewhere, but: Maybe you, who are reading this, is offended by this framing, because you use and enjoy the AI tools. But it's also likely that you, and many other technologists, take moral abdication almost as a point of pride, where the only thing that matters is "capability". In that case, I don't understand the defensive response. Why are you uncomfortable being described as the thing you're bragging to be?

#fuckAI

¹ The stark contrast: The breathless and brainless promotional posts about Glasswing came into my feed at the same time as the posts about the final gutting of CISA. https://www.securityweek.com/white-house-seeks-to-slash-cisa-funding-by-707-million/

Apr 9 at 2:39amWeb
Show threadCindʎ Xiao 🍉1d ago

More details on the cuts at CISA, from @ericjgeller.com.

The budget also proposes a 75% cut to the CyberSentry program, which manages intrusion-detection sensors on the networks of participating critical infrastructure organizations, and a 55% cut to the Cyber Analytic and Data System, which integrates cyber threat data from across the agency. (CISA said technology improvements would allow for less spending on CyberSentry, which it called “a key capability” for its mission.)

Under the spending cuts outlined in the budget, CISA would also retreat from its mission to provide cybersecurity solutions to the rest of the government. The document proposes scaling back CISA’s support to the Justice Department’s Security Operations Center as a Service offering, ending CISA’s support for a General Services Administration program that evaluates cybersecurity vendors’ products and ending the development of a Critical Infrastructure Shared Services program.

https://www.cybersecuritydive.com/news/cisa-trump-budget-fy2027-details/816855/

CISA’s vulnerability scans, field support on chopping block in Trump budget

The president is proposing to shrink the agency by nearly 900 positions.

Cybersecurity Dive
Show threadJon1d ago
@cxiao very well said!
Show threadCindʎ Xiao 🍉1d ago

@jdp23 Thank you Jon!

(I've also been making an effort lately to write down my thoughts after I read something, to think through it more clearly. I'm hoping this will improve my writing a bit, too.)

Show threadFaye21h ago

@cxiao @nuintari if you’re on fedi then you’re probably already reacting to the enclosure of the commons of communication and thought by capital, AKA all the big social networks. It is, after all, the only set of modern social media technologies with resisting enclosure as part of the design goals.

It’s not a great leap to think that people on here would take one look at LLMs and go “ey-up, we’ve seen this exact shit before, fuck off please”.

It’s, sadly, also not a great leap to imagine that anyone who’s fine with “normal” social media doesn’t understand why it’s an issue and so starts to throw accusations when they’re not allowed to undermine a core principle of the culture.

Show threadFaye21h ago
@cxiao @nuintari “enclosing commons” is not an idea that you’re allowed to have in the mainstream
Show threadbotantony21h ago
@cxiao AI-bros love to talk about LLMs "democratizing" software engineering and how you don't need to know programming when they actually mean "tie down your skills and thinking to an LLM provider and hope they are not going to screw you up"
Show threadBill17h ago
@cxiao Wow, that was a hell of a post. I don't think I've ever seen a footnote on a Mastodon post before. That's seriously impressive!
Show threadCindʎ Xiao 🍉8h ago

@Sempf Thanks Bill. I appreciate that on this instance we have the longer character limit, too!

For footnotes, I got the idea from reading @hovav's posts; I think he uses them well.

Show threadHovav Shacham6h ago
@cxiao @Sempf Thanks! I learned how to footnote from @marasawr .
Show threadCindʎ Xiao 🍉6h ago
@hovav @Sempf @marasawr Oh awesome! The art of the footnote is being spread far and wide 🙏