I am still hopeful that we are just a couple of malicious lines of code away from rendering this whole AI abomination inoperable, or at least so heavily crippled with various security measures that it is practically unusable. One can dream, I guess ……
They still haven’t found any viable solution to prompt injection. Every time they put an AI agent in charge of anything important, that’s a huge security risk.
The AIs are very stupid, and it’s trivially easy to trick them into violating security protocols.
I am still hopeful that we are just a couple of malicious lines of code away from rendering this whole AI abomination inoperable, or at least so heavily crippled with various security measures that it is practically unusable. One can dream, I guess ……
They still haven’t found any viable solution to prompt injection. Every time they put an AI agent in charge of anything important, that’s a huge security risk.
The AIs are very stupid, and it’s trivially easy to trick them into violating security protocols.
EU is doing that by putting “you are liable for your AI” in law code. Big Tech was not amused.
I am shocked we are fighting back AT ALL. And I don’t expect it to last.
are you sure that’s not just in Germany because of debacle with AI summaries there