Really interesting read Suny. I'm just one person starting out, so the enterprise side isn't my world, but the human reflex you describe scales right down to my level. I hand my own AI context all day without thinking, because it's useful and saying yes is the easy path. The same instinct you're naming in big companies, just smaller. And the planted email prompt injection is a properly unsettling idea. Nobody needs to breach the firewall, they just need the curious agent to read the wrong message.
Thank you, Samantha! I think that's exactly it, the enterprise problem is really just a human problem at scale. We keep saying yes because the tool is useful. And the prompt injection angle is unsettling because the agent doesn't need to be hacked, it just needs to be convinced.
Wow this was an interesting read. I really enjoyed it
Thank you! It means a lot!
Really interesting read Suny. I'm just one person starting out, so the enterprise side isn't my world, but the human reflex you describe scales right down to my level. I hand my own AI context all day without thinking, because it's useful and saying yes is the easy path. The same instinct you're naming in big companies, just smaller. And the planted email prompt injection is a properly unsettling idea. Nobody needs to breach the firewall, they just need the curious agent to read the wrong message.
Thank you, Samantha! I think that's exactly it, the enterprise problem is really just a human problem at scale. We keep saying yes because the tool is useful. And the prompt injection angle is unsettling because the agent doesn't need to be hacked, it just needs to be convinced.