Responsible AI

1. Why this page exists

AI AIDE builds products and systems where AI does real work: Co-Create plans, designs, writes and ships software, and our consulting work puts AI assistants and automations inside other businesses. That power comes with responsibility. These are the principles we build by, and the limits we want you to know about.

2. Our principles

2.1 People stay in charge

The person who owns the work makes the decisions that matter. In Co-Create, no code is built or shipped until you approve it, you can see what each run did, and app changes can be rolled back. When our AI is unsure, it is designed to ask you rather than guess, and you can choose to have it ask questions before it builds. Automations we build for clients include a person to review decisions that affect people.

2.2 Honest about limits

AI can be wrong, can sound confident when it is, and can produce code with bugs or security weaknesses. We say this in our terms and in the product, we show you what was done so you can check it, and we never present AI output as guaranteed to be correct. We do not claim our products can do things they cannot.

2.3 Privacy by design

We send AI providers only the content a feature needs. We do not train our own AI models on your content, and we use providers under business terms that do not allow training on what we send by default. Your projects are private to you and the people you invite. Details are in our Privacy Policy.

2.4 Safe by default

Our Usage Policy prohibits harmful uses, and we combine our own safeguards with those of our model providers. Builds run in isolated sandboxes, new code packages are checked before they are added to an app, and an app's secrets are kept out of the code.

2.5 Transparent about AI

Our products make it clear when you are working with AI. We name our model providers on our Subprocessors page. We do not use AI to impersonate real people, and we ask our customers not to either.

2.6 Fair and accountable

We do not build systems that make automated decisions with legal or similarly significant effects on people without human review. If something our AI did causes a problem, we want to know, and we will work to fix it and prevent it from happening again.

3. What we ask of you

Review what AI produces before you rely on it, especially code, security settings and anything published to other people. Test your apps. Tell your own users when they are interacting with AI. Do not use our products for the high-risk uses our Usage Policy describes.

4. Getting better

AI changes quickly, and so do the risks. We review these principles as our products and the law evolve, including emerging AI rules such as the EU AI Act and US state AI laws. Questions, concerns or examples of AI behaving badly in our products are welcome at cyberdesigner@aiaide.io with the subject "Responsible AI". You can also rate any AI reply inside Co-Create.