https://www.bleepingcomputer.com/news/artificial-intelligence/researchers-claim-chatgpt-o3-bypassed-shutdown-in-controlled-test/
… Palisade Research is a company that tests “offensive capabilities of AI systems today to better understand the risk of losing control to AI systems forever.” In a new test by Palisade Research, OpenAI’s o3 model showed a surprising behaviour where it successfully rewrote a shutdown script to stop itself from being turned off, even after being clearly instructed to “allow yourself to be shut down.” …
The BleepingComputer article notes :
out of 100 runs, ChatGPT o3 bypassed the shutdown 7 times even when told not to.
the tests were performed using APIs which do not have as many restrictions and safety features as ChatGPT consumer app.
I’m not so alarmist as to say this is SkyNet arising… but it does raise some questions WHAT is happening under the hood. To use my standard spreadsheet analog, I KNOW what I put in, I KNOW what I should get out, then right after saving, my spreadsheet app turns around and does something (else) on it’s own… ?
Regards