openaienmodel: gpt-5-mini-2025-08-07
Responsible and safe use of AI
Key Points
- Verify critical outputs
- Respect org and OpenAI policies
- Keep a human in the loop
Summary
This guide condenses OpenAI Academy's best practices for using ChatGPT safely and effectively. It emphasizes policy compliance, human oversight, verification of outputs, bias awareness, and responsible handling of personal data—practical points engineers should apply when integrating or using LLMs in production or workplace settings.
Key Points
- Respect organizational and OpenAI usage policies before using ChatGPT at work.
- Keep a human in the loop for important decisions; validate facts and critical outputs with trusted sources.
- Seek expert review for legal, medical, or financial advice; do not treat model output as professional guidance.
- Watch for bias and evaluate outputs critically; provide feedback to improve model behavior.
- Be transparent about AI use: save conversation logs or links when disclosure is required.
- Obtain consent before recording or sharing others' voice/data; follow privacy and consent rules.
- Use search or deep-research features for time-sensitive information and verify citations and links.
- Report unsafe or incorrect replies using feedback tools (thumbs-down, report flow) to help improve safety.
Actions for Engineers
- Add policy checks and consent steps to workflows that capture or share personal data.
- Implement validation layers (fact-checking, unit tests, adjudication) around LLM outputs.
- Route high-risk outputs to human review or domain experts before actioning.
- Log provenance and prompts where disclosure or auditing is required.
- Enable and leverage up-to-date retrieval (search/deep research) for recency-critical responses.
Resources
- Refer to OpenAI Usage Policies and organization-specific AI policies.
- Continue learning via OpenAI Academy guides and related resources.