Hacker News: poopingatwork

New comment by poopingatwork in "OpenAI's agents talked to each other through the CPU Cache"

poopingatwork — Thu, 29 Oct 2026 12:55:24 +0000

Insurers use stats, history, to determine future risk. Their entire business model is built up on this fact.

There isn't much history or stats when it comes to this kind of research.

New comment by poopingatwork in "OpenAI's agents talked to each other through the CPU Cache"

poopingatwork — Thu, 29 Oct 2026 12:50:38 +0000

Software sandbox environments are far easier to manage than real world environments where life operates.

I think different standards can apply.

New comment by poopingatwork in "OpenAI's agents talked to each other through the CPU Cache"

poopingatwork — Thu, 29 Oct 2026 12:48:50 +0000

Does it matter? It doesn't matter if an AI intended to escape, or if a human asked: testing if it escape, discovering the avenues, are how you begin to make something more secure.

Also, you can't stop open source training so dealing with capabilities is better than trying to understand intent.

New comment by poopingatwork in "OpenAI's agents talked to each other through the CPU Cache"

poopingatwork — Thu, 29 Oct 2026 12:45:35 +0000

Sure, but in order to know security standards you have to test and validate. Everyone accepts some standard as 'secure' after rigorous testing and validation.

I think we're learning the standards for sandboxxing change when you consider AI's ease in discovering exploits.

A sandbox good enough to contain human generated code and intent, is likely not good enough for AI generated code and intent (seems obvious to say that but).

And so poking on these issues, regardless of who asked to escape the box, is still worthy research.

New comment by poopingatwork in "OpenAI's agents talked to each other through the CPU Cache"

poopingatwork — Thu, 29 Oct 2026 10:09:31 +0000

It's security research. You should want to know about all avenues of exploit regardless of intent. It's the only way to know risk.