Yea the action being presented is trying to bypass the safety guards, but the meme is a judgmental stare. Even if you want to be polite and call it a questioning stare, chatgpt doesn't do any of that itself, either.
LLMs have no ability to think or question or judge at all. It's just correlation of words, and if some naughty words show up in the correlation, OpenAI has programmed other nanny processes to intervene and spit out other output.
The meme is showing an emotional expression, but chatgpt's responses are about as emotional as a greyed out "OK" button on a form that takes about 20,000,000,000,000,000 times the processing power.