An interesting but frustrating read. I agree with the complaints, but lament the lack of concrete plans to fix any of it. One wonders if we’re all waiting for someone else to create the vision.

Edit: here’s one: https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf

I think they threw it behind a login wall because it was getting hit so hard.

Original url: https://nanda.media.mit.edu/ai_report_2025.pdf

All instances of the link in public either reroute to the main NANDA page or have been replaced by the access form to request access to the research.

I think that lead author Aditya Challapally may still have an industry job and I wonder if his employer (M$) objected.

security researchers have repeatedly demonstrated that implementing so-called “lawful” backdoors is inherently flawed as such vulnerabilities would inevitably be discovered, accessed, and exploited by cybercriminals and black-hat hackers.

Yes, that. Every time this comes up, it requires a rehashing of just how dramatically bad it is as a practice. There’s no such thing as a “back door only for the good guys”

Notably, the European Commission makes no mention of new partnership initiatives with the United States.

Probably wise.

deleted by creator