It’s not an agent, it’s whoever used whatever piece of software in whatever which way to connect to those systems, that’s who is responsible and both them and the people who set up an insecure Medicare network should be in jail. For a long time.
accountability is needed here; heads need to roll
So who is going to jail now? I am sure a company is responsible for its actions, right?
This is a big problem with AI, laws have not remotely caught up to these scenarios. Then you have Argentina creating legal structures for AI corporations that could essentially do this type of stuff, but evade prosecution because the corporation will give the owner limited liability and with no employees, the only punishment can be to shut down the corporation. Then they can spin up a new AI corporation and do it all over again in theory. Technology is advancing far beyond our ability to control it properly and tech bros don’t care much about consequences since they have historically been so minimal. Move fast, break shit is a psychopathic ethos.
Its easy. There are laws against hacking. If you used a script its still you who go in jail. If you used an Ai its still you who go to jail.
Do not let them try to spin this into the autonomy narrative.
Someone set up the agent, somebody gave it the prompt that caused this to happen. Some one did not do their job properly.
Ai agents are not independent actors, they are tools.
If I missmanage a forklift, I am held responsible. Breaks failed? I failed to do the right maintenance. Load fell down? I did not secure it right.
This is not some impossible gordian knot to solve…
Reading the story it’s not clear who set up the agent. It’s an agent running on OpenAI but it’s not clear whether they’re saying a customer instructed an agent to hack the Australian medical board and got around guardrails to prevent that, or that they’re saying their own internal work agent decided to hack the Australian medical board at random.
In both cases OpenAI has culpability. In the latter case, they should be held fully responsible.
It would be AWFUL if someone used AI to hack the Department of Home Affairs and grant all 417 visa applications. Just saying.
Edit: Maybe only the applications that are 100% decision ready and have been lodged on the 5th of July? No? Ok.
