AI Transparency and Limitations

Quantx includes AI-generated responses and, where available, tools that help carry out tasks. You are interacting with an automated system unless the interface explicitly identifies a human. This page explains important limitations and should be read with the Terms of Service and Privacy Policy.

1. Responses require judgment

AI can produce plausible but incorrect facts, faulty calculations, unsafe code, fabricated citations, biased statements, or incomplete advice. Responses may not reflect recent events or your full circumstances. Review material before relying on it, executing it, or sending it to others. A confident tone or a model's explanation is not evidence that its answer has been verified.

Search links and citations identify external material where available, but they do not guarantee that a source supports every generated statement. Review the actual source. External pages can change after retrieval. We do not guarantee comprehensive coverage of the web or all relevant evidence.

2. Models and processing locations

Model availability and performance vary with device memory, runtime support, model version, task size, and network conditions. Some work may run on your device; other work may use connected nodes or external providers under the applicable disclosures. The interface identifies processing using the mode and status labels shown in the chat and search interface. We do not equate the number of enrolled devices with capacity available for your task.

An offline feature can use only the models and content actually available offline. Live web search, remote nodes, synchronization, account checks, and updates may require connectivity. Network participation does not automatically combine distributed memory into one accelerator or guarantee parity with a centralized model.

3. Actions outside a conversation

Where an agent can use a connector, file, application, or account, review the granted scope. Instructions, retrieved pages, and documents can contain misleading directions. Treat unexpected requests to disclose secrets, send money, grant permissions, or change security settings with caution.

The feature must explain which actions are automatic and which require confirmation. An approval to draft a message differs from approval to send it. An approval to inspect a file differs from approval to delete it. You remain responsible for authorizing actions within your authority. A stop request cannot guarantee reversal of an external action already completed.

4. Professional and consequential decisions

Quantx is not a substitute for a licensed professional or emergency service. Consult an appropriate professional for medical, legal, financial, and other consequential decisions. Do not use AI Output as the sole basis for a decision about a person's rights, eligibility, safety, or access to essential opportunities. Required human review and other legal safeguards cannot be replaced by a disclaimer.

5. Privacy and confidential material

Before submitting sensitive material, check the processing route and your authority to disclose it. A locally installed app can still use remote services. Encryption in transit does not establish that an executing node cannot access the information it processes. A job split into smaller parts is not automatically anonymous.

The Privacy Policy identifies actual data practices, training uses, human access, and retention. We do not promise that every prompt is private from all recipients or that confidential data is suitable for every enabled route.

6. Output rights and disclosure

Your rights in Input and Output are described in the Terms of Service. AI-generated material may not qualify for exclusive intellectual-property protection, may resemble material produced for others, and may require further rights review. Do not present an output as an authentic recording or statement from another person. Disclose AI involvement where law, professional rules, a platform, or the circumstances require it.

7. Research and environmental statements

Research papers, diagrams, prototypes, and financial or environmental scenarios distinguish measured results from assumptions and proposals. They do not establish that every Quantx workload is cheaper, faster, uses less water, or produces fewer emissions than every data-center alternative.

An environmental comparison depends on comparable task quality, electricity sources, runtime, networking, utilization, hardware manufacturing and lifetime, cooling, and the infrastructure actually displaced. Concept hardware has not demonstrated a product-specific benefit merely because its rendering is compact.

8. Feedback and material changes

Report incorrect, unsafe, or misleading behavior through the thumbs up and thumbs down buttons on each response or support@getquantx.com. Include only information needed to investigate. Feedback content is handled under the Privacy Policy. We will update this notice for material changes in AI functionality or the processing model and provide additional disclosure where required.