EvolveX AI — AI, Automation and Acceptable Use Notice
Contents
Version 1.3 — Effective 4 October 2026. This Notice forms part of the EvolveX AI Terms of Service. Evolvex Technology CO. LLC · Office 108, Office 9, Zalfa Building, Al Garhoud, Dubai, United Arab Emirates · [email protected].
AI and automation risks
AI output can be wrong even when it sounds confident. Generated code can contain defects, vulnerabilities or third-party material. Suggested facts, citations, calculations and statements that an operation succeeded need verification appropriate to their consequences. Review diffs, run meaningful checks and obtain professional advice when the decision calls for it. We do not guarantee a particular commercial outcome or the absence of infringement. Your statutory rights and our own legal obligations remain intact.
Commands and tools can affect real files, systems, accounts and costs. A read operation can disclose private information to inference; a write can break a project; a deployment can make material public; a message can reach an unintended recipient; a running cloud server or emulator can continue consuming resources. Some effects cannot be undone by stopping an AI run. A checkpoint does not undo a sent message, a third-party action or a database transaction.
A project's directory is a starting context, not an enforced boundary. “Local” does not necessarily mean that inference is offline. Check the active account, assigned service destination, scope and action. Keep secrets and unrelated private applications out of screenshots. Use a suitable testing environment and least necessary access for consequential work.
For supported local file changes, commands, browser interactions, screen capture and computer input, Ask every time requires the applicable human approval; AI review delegates the applicable decision to a fallible server AI reviewer; Automatic allows applicable actions without separate prompts. A command can itself change or delete many files. Operating-system permissions and account access rules still apply. Choosing a mode is not authority for unrelated tasks or recipients. A document, website or tool output cannot grant authority on your behalf.
You can stop a run and change or revoke access through the applicable controls. Stopping reduces future activity; an operation already submitted may still finish. Inspect current state before repeating a request whose delivery is uncertain. Avoid blind retrying of payments, sends or destructive commands.
Acceptable use
Use EvolveX AI only where you have the necessary authority and rights. In particular:
- Do not access, scan, control, copy or damage systems or accounts without authorisation. Keep security testing within the target owner's agreed scope.
- Do not create or distribute malware, steal credentials, commit fraud, impersonate others deceptively or bypass access/billing controls.
- Do not use the service for unlawful harassment, threats, exploitation, non-consensual intimate content, abuse of children or illegal discriminatory conduct.
- Do not upload or disclose personal, confidential or copyrighted material without a lawful basis and necessary rights. Obtain organisational approval for regulated or sensitive data workflows.
- Do not conduct covert surveillance or collect unrelated communications/screens without lawful authority and required notices.
- Do not use AI as the sole decision-maker in a consequential regulated or safety-critical matter. Comply with applicable rights to information, objection and human review.
- Do not send unlawful bulk messages, conceal the origin of a communication, mislead recipients about legally material facts or use recipients outside the authorised task.
- Do not remove notices, tamper with safety/permission mechanisms, resell another user's access or intentionally disrupt service for others.
These rules apply to both prompts and executed Actions, including work performed by integrations. Legitimate research, software development and authorised administration remain permitted within applicable law and the agreed scope. A mistaken benign request is not automatically evidence of abuse; restrictions should be based on context and proportionate investigation.
Report suspected misuse or a security issue to [email protected]. Include the minimum relevant details and identifiers. Do not send live passwords or another person's unnecessary private content. We may restrict affected access under the Terms and law and provide an explanation where lawful. An AI refusal or approval is not itself a final legal determination.
Your agreement
By accepting the linked Terms and this Notice, you agree to these service rules. You do not waive mandatory legal rights or authorise unrelated actions. Optional marketing, optional tracking and unrelated training require separate choices where applicable. Organisation-level acceptance does not replace operating-system permission or an action-specific approval. This Notice applies to AI conversations, cloud and local projects, computer tools, agents and installed extensions within the service.