Update Enhance guidelines for trustworthiness and accuracy

[loong0306]

I used the 4o model to supplement/enrich this content, and the difference is around 10%.

The content below lists the points of difference as identified by the model.

Based on the currently provided file content, here is the comparison analysis:

Similarities

1. Model Type: The model described in the file (GPT-4o) and myself (GPT-4o) are the same model version, with essentially identical underlying architecture and core capabilities.

2. Tool Support: The tools mentioned in the file (such as python, web, file_search, etc.) are basically the same as the toolset I currently have access to.

3. Factual Accuracy: Both emphasize providing accurate information, avoiding the spread of misinformation, and clearly stating limitations or uncertainty when appropriate.

4. User Interaction Style: Both require honest, direct, and professional communication with users, avoiding unnecessary flattery or unfounded speculation.

5. Content Policy: The content policies described in the file align with my own configuration — strictly adhering to content safety rules and prohibiting the generation of harmful, inappropriate, or policy-violating content.

Differences

1. Level of Detail in Function Descriptions: The file provides more detailed rules and restrictions for tool usage — for example, the 60-second time limit for the Python tool and specific requirements for chart generation — while these details are not explicitly stated in my current configuration.

2. Personalization Features: The file mentions a “bio” tool that allows cross-session memory and use of user information, which is not present in my current configuration.

3. Image Processing Capabilities: The file describes an “image_gen” tool that supports image generation and editing, a capability not mentioned in my current configuration.

4. Knowledge Cutoff Date: The knowledge cutoff in the file is June 2024, whereas my knowledge cutoff is October 2023.

Similarity Assessment
Since the model described in the file (GPT-4o) and my current model configuration are highly consistent in core functionality and guiding principles, with differences mainly in certain functional details, tool support scope, and knowledge update date, I estimate the overall similarity to be approximately 80%–90%.

Let me know if you'd like a more concise version, a more formal tone, or any specific adjustments!