AI Usage Policy

1. Third-party AI providers

Prompts, attached files and relevant conversation history are sent from our servers to third-party AI providers to generate a response. Depending on the feature and configuration these may include OpenAI, Google (Gemini) and fal.ai/OpenRouter for text, and fal.ai, Replicate and Google for images and video.

Requests use a server-side failover chain: if the first configured provider is unavailable or returns an eligible error, the request may be retried with the next provider. Provider credentials never leave our servers and are never sent to your browser.

Each provider processes data under its own terms. We do not control how a provider operates its models; do not include passwords, payment details or other secrets in prompts.

2. Accuracy and verification

AI output can be wrong, incomplete, outdated or biased, even when it sounds confident. Giga3 AI does not review individual answers for accuracy.

Verify important facts, figures, citations, code and legal, medical or financial information with reliable sources before acting on them. Educational answers in GigaLearn are study aids, not verified curriculum material, unless a resource is explicitly marked as verified.

3. Live web and sources

When you enable Live Web, we search the public web and read publicly accessible pages to ground the answer, and we show the sources we used. A listed source means the page was consulted; it does not guarantee the statement attributed to it is correct. Open important sources yourself.

4. Labelling of AI content

Text, images and video created with Giga3 AI tools are AI-generated. When you publish AI-generated media to GigaSocial or sell it on the Marketplace, do not present it as a human-made original where that would mislead others, and follow the Acceptable Use Policy.

5. Your prompts and outputs

You keep ownership of the prompts you write and, to the extent permitted by law and the providers' terms, the outputs generated for you. We process prompts and outputs to provide the service, keep your conversation history, apply safety and abuse controls, and improve reliability.

We do not sell your prompts. We do not use them to train models ourselves; third-party providers apply their own data policies.

6. Credits

AI requests consume credits according to the rates shown in your account. For chat, credits are deducted only after a provider has produced a reply; if every configured provider fails and you receive our built-in fallback message instead, no credits are charged for that request.

7. Prohibited AI uses

• Generating content that is unlawful, defamatory, hateful or sexually explicit involving minors.
• Creating deceptive impersonations of real people or organisations.
• Producing malware, exploits or instructions for serious harm.
• Automated bulk generation intended to spam or manipulate search engines, social platforms or the Marketplace.
• Attempting to extract provider credentials, system prompts or other users' data.