This is such obvious con artistry, like the old William Castle horror movie ads where they promised to have a nurse in the theater because the film was so scary you might need a medical discharge from watching it.

Does anyone but VCs fall for this shit?

April 09, 2026 - 12:45 UTC

6
0
85
Reporters
1
0
3
ZING
0
0
2
Ref www.tcm.com/articles/179093/homicidal
2
0
8
That "During the last two minutes ..." part was great!
0
0
0
Saw that movie on TV when I was about 10 (around '68-9) late one Friday night in Winnipeg ("The Chiller Thriller"). Messed with my mind and I thought about it for years. Much later I tracked it down (having forgotten the title/sketchy memories) and watched again. One of Castle's better ones!
2
0
2
Generally, I'm with you. But I'm with @ncweaver.skerry-tech.com on this one. As I understand it, one of the things that AIs are actually good at these days is finding computer security vulnerabilities. And vulns are everywhere. So, even if this is largely hype, I'd believe there's a kernel of truth.
0
0
2
I too have developed an AI that is so powerful I will never release it, please give me $500 billion. Do not ask me for evidence, it is not permitted.
0
0
2
I was going to say Kevin Roose will, then I clicked through
0
0
1
There is a fair bit of exaggeration but this is IMO mostly-real: One thing that LLMs are remarkably good at is bug/vulnerability finding, because the "bullshit machine" problem isn't there, the results themselves are self-validating and there is a lot of non-LLM infrastructure involved in this.
0
0
3