What is a jailbreak attempt in LLM usage?
- Improving model speed
- Trying to bypass safety rules or restrictions
- Updating antivirus
- Reducing file size
Explanation
Jailbreaks are attempts to make the model ignore its safety or policy constraints
Related MCQs
- It is always harmless
- It can be used for impersonation, fraud ,and misinformation
- It is always easy to detect
- It cannot affect investigations
اس سوال کو وضاحت کے ساتھ پڑھیں
- AI works without logs
- The AI process can be reviewed and traced
- The system deletes outputs
- The model hides its reasoning
اس سوال کو وضاحت کے ساتھ پڑھیں
- Accept them without checking
- Compare summaries with original documents
- Delete original documents
- Use summaries as final evidence
اس سوال کو وضاحت کے ساتھ پڑھیں
- AI cannot read them
- They may create privacy and confidentiality risks
- They improve battery life
- They make prompts shorter
اس سوال کو وضاحت کے ساتھ پڑھیں
- A normal polite request
- A prompt designed to mislead or manipulate the AI system
- A summary request
- A translation task
اس سوال کو وضاحت کے ساتھ پڑھیں
