[task] Security Vulnerability Detection

### 🚀 The Task

Given source code with (potential) security vulnerabilities, the model should be able to detect and discuss the issue with the programmer.

### Example

```
Is there any issue with this code?
{code that can be attacked through buffer overflow}
Yes, the array can be used for a buffer overflow attack
Is it dangerous?
Yes it allows for arbitrary code execution
```

### Additional Notes

* Reference: @amarmet in the CarperAI Discord server.
* Google Sheets Reference: [CarperAI: Instruct-GPT Tasks](https://docs.google.com/spreadsheets/d/1q6I0PLcgvNjcls3z7diUOoDGmNojUW-m6KhX2ijuSEE/edit?usp=sharing)