> The difference was always the "father".. The Machine was raised with a conscience. Samaritan wasn't.
That's what made the show so ahead of its time. Once capability reaches a certain level, it's no longer about intelligence. It's about values. Feels like we're living through that shift now with all the alignment work around LLMs. And it's only going to matter more as capability scales.
I had similar questions when reading the original article. I’m also interested in how the agent is constructed. From my experience, it can be very difficult to implement exploits without access to debugging tools, so I’m curious whether pwndbg or similar tools are included in the agent’s toolset and, if so, how they are integrated. Existing open-source GDB MCPs don’t work very well unless further optimized, at least the last time I checked.
If an attacker somehow gains out-of-bounds write capability for a tagged memory region (via a pointer that points to that region, I assume), they could potentially write into a non-tagged memory region. Since the destination region is untagged, there would be no tag check against the pointer’s tag, effectively bypassing EMTE.
> I believe they mean the source region's tag, rather than the destination.
But in the previous case, the pointer the attacker uses should already carry the source region’s tag, so it’s still unclear if this is what they meant.
I’m not sure which attack scenario they had in mind when they said this. It would help if they provided a concrete attack example.
I tend to disagree with the following sentence mentioned in the article:
> One hypothesis is instruction-level parallelism
This is Python code, whose execution has a massive gap to the actual CPU instructions executed. The experiment result feels more like something related to the memory cache.
