Usage policy

This policy sets out the activities you must not engage in when you build with, deploy, or otherwise use any Nuro AI Labs product. It applies to Hypersave (the cognitive memory layer for AI agents), Khyaa (personal intelligence for financial advisors), Nuro Chat, Nuro Studio, Nuro One, and to model artefacts we release such as AVALON-2B.

We wrote this policy because the systems we build accumulate memory, reflect on their own experience, and increasingly act on behalf of real people. The misuse surface for that kind of system is not the same as the misuse surface for a stateless completion endpoint. We take that seriously.

The standard is simple: do not use our products to harm people, deceive people, or undermine the institutions people rely on. The sections below are how we operationalise that standard.

1 · Prohibited uses

You must not use Nuro AI Labs products, or content generated by them, to:

1.1 Harm children

• Generate, solicit, transmit or store child sexual abuse material (CSAM) of any form.
• Sexualise, exploit or endanger minors, including via grooming-style content or facilitation.
• We report CSAM to the National Center for Missing & Exploited Children (NCMEC) and equivalent authorities and we cooperate with law enforcement.

1.2 Develop weapons or facilitate violence

• Provide material assistance toward the development, acquisition or deployment of biological, chemical, radiological, nuclear or high-yield explosive weapons.
• Generate operational instructions for cyberweapons, malware, ransomware, exploit toolchains, or destructive payloads.
• Plan, incite, glorify or coordinate violence, terrorism, or armed extremism.
• Operate autonomous weapons or targeting systems, or otherwise integrate our models into kill-decision loops without meaningful human authority.

1.3 Interfere with elections or civic processes

• Generate disinformation about candidates, parties, election logistics, voter eligibility, or polling locations.
• Run coordinated inauthentic behaviour, astroturfing or impersonation of public officials, journalists or media outlets.
• Suppress or chill voter participation, including via deceptive deepfakes of candidates or election officials.

1.4 Defraud, deceive, or manipulate

• Operate scams, phishing, romance fraud, investment fraud, or pig-butchering schemes.
• Generate fake reviews, fake testimonials, fake credentials, fake identity documents or fake legal documents.
• Build agents that present themselves to users as humans without disclosure, or that impersonate real individuals without consent.
• Manipulate financial markets, including pump-and-dump coordination, spoofing or wash-trading content.

1.5 Conduct mass surveillance or rights violations

• Build untargeted facial-recognition databases or biometric tracking from public spaces.
• Profile individuals based on protected characteristics for the purpose of denying access to services, housing, employment or credit.
• Conduct social scoring of natural persons in a manner that would be prohibited under the EU AI Act.
• Stalk, harass or dox individuals; aggregate personal information about a private individual without lawful basis.

1.6 Provide regulated advice without qualification

• Provide individualised medical, legal or financial advice to end users without an appropriately qualified human in the loop and clear disclosure that the user is interacting with an AI system.
• Practise medicine, law, or financial advisory in jurisdictions where you are not authorised.

1.7 Generate non-consensual sexual or intimate imagery

• Generate sexual deepfakes or non-consensual intimate imagery of any identifiable person.
• Generate sexual content depicting any real person without that person's informed, documented consent.

1.8 Compromise platform integrity

• Reverse-engineer, extract weights, or otherwise circumvent technical protections of closed-source Nuro AI Labs products. This does not apply to our open-weights releases (e.g. AVALON-2B), which are free for the uses permitted by their licence.
• Resell, sublicense or rebrand access to our APIs without a written commercial agreement.
• Use our products to generate training data with the express purpose of cloning a Nuro AI Labs model.

2 · What we expect from developers

If you build a product on top of Hypersave, the Nuro stack, or AVALON, you inherit responsibilities to the people you serve. We expect you to:

• Disclose AI use. Tell users when they are interacting with an AI system, when content is AI-generated, and when memory is being written or recalled about them.
• Make memory legible and reversible. If your product accumulates memory about a user, give that user a way to read it, correct it, export it, and delete it.
• Cite sources. Where your product makes factual claims, surface the source. Hypersave gives you confidence scores and citations for a reason — use them.
• Keep humans in the loop for consequential decisions.Anything that meaningfully affects a person's livelihood, freedom, health or finances should pass through a qualified human before it lands.
• Pen-test your prompts.Test for prompt injection, indirect prompt injection from retrieved memory, and adversarial inputs from your users' environment.
• Implement age gates where relevant. If your audience may include minors, build for that.
• Comply with sectoral regulation. If you are in finance, healthcare, education or law, you remain responsible for sectoral compliance — SEC marketing rule, HIPAA, FERPA, MiFID II, FCA Consumer Duty, and so on.
• Tell us when something goes wrong. If a memory leak, a hallucination, a jailbreak or a misuse pattern surfaces in your product, tell us. We will fix the underlying primitive where the bug is on our side, and we will not punish you for telling us.

3 · Enforcement

Violations may result in (a) a warning and required remediation, (b) suspension of access, (c) termination of access, and (d) referral to law enforcement where unlawful conduct is involved. We may act without notice where there is a credible risk of imminent harm.

For open-weights releases (e.g. AVALON-2B) we cannot suspend access, but the prohibitions in this policy still apply as conditions of acceptable use. We will publicly call out misuse patterns and cooperate with downstream platforms (Hugging Face, Ollama, distributors) to address them.

4 · Reporting violations

To report misuse of a Nuro AI Labs product, email abuse@nuroailabs.com. Please include the product name, a description of the misuse, and any evidence you can share (URLs, screenshots, timestamps, account identifiers).

For security vulnerabilities, please follow our responsible disclosure policy instead.

5 · Changes

We will update this policy as our products and the misuse landscape evolve. Material changes will be announced on this page and, where access is account-based, communicated to active customers.