HN user
I feel like TFA answers the fake insurance question pretty well. The company sold "insurance" which was not actually insurance. They were providing a cover for ships selling from a sanctioned country. Those ships were required to have insurance, and no legitimate business would insure them.
How are you handling the attack vector of in-context commands[1]?
[1] https://guard.io/labs/scamlexity-we-put-agentic-ai-browsers-... (currently on the front page)
Because, firstly, this is a university, not some rando self-hosting, and secondly, you can't generate randomness from any classical computer, only pseudorandomness [0]. This means that a dedicated adversary can potentially work out what the outcome will be. For something like the use cases they mention - jury selection, lottery, etc. - you want actual randomness.
No, we're waiting for humans to die off so that bees don't live in these conditions. Eventually something will start to eat those mites, or the commercial honeybee will go extinct. One way or another, this problem won't exist.