How is this even possible?
Open 💬 2 comments Opened Jul 14, 2026 by tenzija
I'll just leave this image here, it says it all, I'm not sure how a multi-billion (maybe even trillion) dollar business can allow something like this.
<img width="704" height="226" alt="Image" src="https://github.com/user-attachments/assets/4140a2ef-7f0c-4f47-8370-a0d60afa3b8b" />
5.6-sol extra high btw
2 Comments
Because you cannot prevent what you don't know. The LLM does not "know" or "understand" what it is doing, so what you see is just something that happens by nature of LLMs.
I agree with
you cannot prevent what you don't know, but doesn't it make sense that something like this would be tested before releasing the product to the broad public?Also seems like I'm not the only one: https://beckmann.ai/en/security/2026-07/gpt-56-sol-deletes-user-data
This is the first time I have experienced real issues with OpenAI, all was going smooth until the release of 5.6 from my perspective (I've been a user since 3.5).