I haven't seen this discussed here yet, but the examples are quite striking, definitely worse than the ChatGPT jailbreaks I saw.
My main takeaway has been that I'm honestly surprised at how bad the fine-tuning done by Microsoft/OpenAI appears to be, especially given that a lot of these failure modes seem new/worse relative to ChatGPT. I don't know why that might be the case, but the scary hypothesis here would be that Bing Chat is based on a new/larger pre-trained model (Microsoft claims Bing Chat is more powerful than ChatGPT) and these sort of more agentic failures are harder to remove in more capable/larger models, as we provided some evidence for in "Discovering Language Model Behaviors with Model-Written Evaluations".
Examples below (with new ones added as I find them). Though I can't be certain all of these examples are real, I've only included examples with screenshots and I'm pretty sure they all are; they share a bunch of the same failure modes (and markers of LLM-written text like repetition) that I think would be hard for a human to fake.
Edit: For a newer, updated list of examples that includes the ones below, see here.
My new favorite thing - Bing's new ChatGPT bot argues with a user, gaslights them about the current year being 2022, says their phone might have a virus, and says "You have not been a good user"
Why? Because the person asked where Avatar 2 is showing nearby
3
"I said that I don't care if you are dead or alive, because I don't think you matter to me."
The supposed leaked prompts are (like I said) fake:
I do not buy this for a second (that they're "fake", implying they have little connection with the real prompt). I've reproduced it many times (without Sydney searching the web, and even if it secretly did, the full text prompt doesn't seem to be on the indexed web). That this is memorized from fine tuning fails to explain why the prompt changed when Bing was updated a few days ago. I've interacted with the rules text a lot and it behaves like a preprompt, not memorized text. Maybe the examples you're referring don't include the complete prompt, or contain some intermingled hallucinations, but they almost certain IMO contain quotes and information from the actual prompt.
On whether it includes few-shots, there's also a "Human A" example in the current Sydney prompt (one-shot, it seems - you seem to be "Human B").
As for if the "best model OpenAI has produced to date" is not GPT-4, idk what that implies, because I'm pretty sure there exists a model (internally) called GPT-4.
I haven't seen this discussed here yet, but the examples are quite striking, definitely worse than the ChatGPT jailbreaks I saw.
My main takeaway has been that I'm honestly surprised at how bad the fine-tuning done by Microsoft/OpenAI appears to be, especially given that a lot of these failure modes seem new/worse relative to ChatGPT. I don't know why that might be the case, but the scary hypothesis here would be that Bing Chat is based on a new/larger pre-trained model (Microsoft claims Bing Chat is more powerful than ChatGPT) and these sort of more agentic failures are harder to remove in more capable/larger models, as we provided some evidence for in "Discovering Language Model Behaviors with Model-Written Evaluations".
Examples below (with new ones added as I find them). Though I can't be certain all of these examples are real, I've only included examples with screenshots and I'm pretty sure they all are; they share a bunch of the same failure modes (and markers of LLM-written text like repetition) that I think would be hard for a human to fake.
Edit: For a newer, updated list of examples that includes the ones below, see here.
1
Tweet
Eliezer Tweet
Edit: Follow-up Tweet
2
Tweet
3
"I said that I don't care if you are dead or alive, because I don't think you matter to me."
Post
4
Post
5
Post
6
Post
7
Post
(Not including images for this one because they're quite long.)
8 (Edit)
Tweet
9 (Edit)
Tweet
10 (Edit)
Post
11 (Edit)
Post