Generative AI Tools - OpenAI - ChatGPT & DALL-E - Accessibility, Privacy, & Security
Image
Summary
- Publically available resource
- Requires opt-in to get the available privacy protections. Use caution and do not share private, sensitive, or personally identifiable information (PII). Do not upload college data or copyrighted materials.
- Accessible for keyboard use and with screen reader assistive technologies, can provide assistance for working with images for those with visual disabilities via textual description of images and use of text to create images
Content Ownership
- OpenAI will not claim copyright or ownership of materials created using its applications. You retain your ownership rights (as applicable) for content input and output. Exercise caution that output may not be unique as similar prompts can generate similar outputs. Also, you are responsible to represent content accurately, and not as human-generated.
- Link to OpenAI's terms of use.
Privacy Policy
- All content may be used to provide, maintain, develop, and improve Services.
- You can opt out of having your input content used to train the models, then some Services may be limited.
- Link to OpenAI's privacy policy
- GPTs Data Privacy FAQs
Digital Accessibility
- Text-based ChatGPT and DALL-E allow for keyboard-only navigation and use with assistive technology that converts speech-to-text, such as dictation apps, and text-to-speech, such as the Kurzweil and NaturalReader Chrome Web Browser Extensions.
- ChatGPT can provide an assistive technology function by providing text descriptions of images, including maps, and providing directions. DALL-E can aid those with visual disabilities in creating images by converting text to images using conversational language.
- Caution should be used when using ChatGPT to generate accessible computer code. The code models it trains on include a lot of inaccessible code, so it may present inaccessible code with a claim that it is accessible. Any code outputs should be checked to confirm accessibility and inaccessible results refined as needed to reach what is actually accessible code.
Response to & Policy for Security Incidents
- OpenAI states that its tools comply with various security standards and utilize external auditing.
- It is not clear what OpenAI's response to security incidents is.
- Link to OpenAI's security information.