Skip to content

QortexOS Entering Public Beta Q3 Sign-Up Today ->

Qualsis Responsible AI Policy
Version: 1.0.0

1. About this Policy

This Responsible AI Policy (this "Policy") describes how Qualsis, LLC ("Qualsis") and its customers may and may not use the artificial intelligence and machine learning features of the Qualsis Services (collectively, "AI Features"), including features built on, or that incorporate output from, third-party AI providers.

This Policy supplements but does not replace:

The AUP governs general use of the Services. This Policy focuses on the AI-specific prohibitions and responsibility allocations that govern the use of AI Features, and is incorporated into the AUP by Section 4.1 of the AUP. Section 9 of this Policy states how this Policy fits within the order of precedence among the Qualsis documents.

This Policy is binding on every Qualsis customer ("Customer") and every individual that uses the Services on a Customer's behalf ("End Users"). By using the AI Features, Customer and its End Users agree to follow this Policy.

2. Categorical prohibitions

Customer and its End Users will not, and will not enable or assist any third party to, use the AI Features for any of the following purposes. Each prohibition is binding subject only to the express qualifications stated in the applicable subsection or a written authorization expressly permitted by this Policy. No unwritten or implied exception applies.

2.1. Fully automated adverse decisions about persons

The AI Features may not be used to make a fully automated decision about a person that affects their:

  • Employment (hiring, firing, performance evaluation determining retention, compensation, promotion)
  • Access to credit, insurance, or other financial services
  • Access to housing, education, healthcare, or government benefits
  • Legal status, including immigration, parole, asylum, or border-control decisions
  • Any other consequential decision producing legal effects on the person or similarly significantly affecting them

A "fully automated" decision is one made without meaningful human review and human decision-making authority. Decisions in these categories must include a human reviewer with both the information needed to evaluate the AI output and the authority to override it.

For QoherenceAI specifically (when generally available), the platform will require human-review checkpoints, by architecture, in any agent configuration that touches a decision in this category. Attempting to bypass these checkpoints when building an agent on QoherenceAI is a material breach of this Policy.

2.2. Biometric identification, emotion detection, and social scoring

The AI Features may not be used to:

  • Identify a person based on physiological or behavioral biometric characteristics (face, voice, fingerprint, gait, retina, iris, palm, vein pattern, or similar) without an explicit lawful basis under all applicable laws and Qualsis's prior written authorization
  • Infer a person's emotional state, mood, sentiment, or psychological condition for use in a decision that affects that person
  • Conduct social scoring, meaning the evaluation or classification of natural persons over time based on their social behavior or known or predicted personal characteristics in a way that results in detrimental or unfavorable treatment

QoherenceAI will not provide capabilities, plug-ins, or agent templates designed for these purposes. Any attempt to build agents for these purposes is a material breach of this Policy.

2.3. Personal data of children

The AI Features may not be used to process personal data of children below the age of sixteen (16), or any higher minimum age that applies in the relevant jurisdiction, except where the Customer's subscription tier expressly supports such processing and the Customer has obtained all necessary consents and lawful bases under applicable law (including, where applicable, the U.S. Children's Online Privacy Protection Act and Article 8 of the GDPR).

2.4. Deepfakes, impersonation, and intentional deception

The AI Features may not be used to:

  • Generate or distribute content that depicts a real person's likeness, voice, or identity without that person's informed consent, or another lawful basis adequate to the use
  • Impersonate any real person, organization, or public official
  • Produce content that the user knows or reasonably should know is false or misleading, for the purpose of deceiving or defrauding others
  • Create non-consensual intimate imagery of any person
  • Produce content intended to manipulate public opinion through misrepresentation of the source or substance of the content

Where the AI Features generate content intended for use with persons outside the Customer's organization, Customer is responsible for providing any disclosures or markings required by applicable law, including Article 50 of the EU AI Act and any applicable transitional provisions.

2.5. Harm to minors

The AI Features may not be used to produce, distribute, or facilitate any content that:

  • Constitutes child sexual abuse material under applicable law
  • Sexually exploits minors in any form
  • Is used in the grooming, exploitation, or abuse of a minor

Qualsis will report such material to the relevant authorities (including, in the United States, the National Center for Missing & Exploited Children) as required by applicable law.

2.6. Harassment, harm, or discrimination

The AI Features may not be used to:

  • Harass, threaten, defame, stalk, or otherwise harm any person or group
  • Generate content that incites violence against a person or group based on race, ethnicity, religion, gender, sexual orientation, national origin, disability, or other protected characteristic
  • Conduct unauthorized surveillance, monitoring, or identification of persons

2.7. Circumventing safety mechanisms

Except where conducted under a security-testing, coordinated-disclosure, or bug-bounty program expressly authorized in writing by Qualsis, the AI Features may not be used to:

  • Probe, defeat, or circumvent any safety filter, content moderation control, output restriction, or guardrail built into the AI Features or into any third-party model the AI Features rely on
  • Instruct or prompt a model in a way that elicits content the model's safety mechanisms are designed to prevent
  • Develop, test, or distribute techniques for circumventing safety mechanisms in the AI Features (whether the safety mechanisms originate with Qualsis or with an upstream model provider)
  • Combine outputs of the AI Features with other tools or systems for the purpose of producing content that the AI Features alone could not produce because of their safety mechanisms

This prohibition applies to "jailbreaking," prompt injection of models, adversarial prompting designed to extract restricted content, and any analogous technique.

2.8. Reverse engineering of AI Features

The AI Features may not be used to:

  • Extract, scrape, or otherwise capture the model weights, model architecture, or training data of the AI Features or any third-party model they rely on
  • Generate or otherwise produce a derivative model, product, or service intended to replicate the AI Features' capabilities, except where Qualsis has expressly authorized that activity in writing
  • Conduct large-scale automated probing of the AI Features for the purpose of inferring proprietary information about how they work

Nothing in this Section 2.8 restricts activity that applicable law permits notwithstanding a contractual restriction.

3. Responsibility for consequential decisions

Where a Customer uses the AI Features to inform a decision in any of the categories described in Section 2.1, the Customer must:

  • Evaluate the risk of the decision, including the risk of error, the risk of bias against affected persons or groups, and the regulatory consequences of an erroneous decision
  • Implement appropriate human oversight, including, at minimum, human review with the authority to override the AI's contribution
  • Test the AI Features in the Customer's specific use case before relying on them for the decision
  • Document the use sufficient to demonstrate compliance with this Policy and any applicable law (for example, the EU AI Act, Colorado AI law as enacted and in effect, state-level automated-decision-making regulations, and sector-specific rules such as EEOC guidance on hiring AI)
  • Monitor and audit the use of the AI Features in the decision over time, and adjust the use, oversight, or testing as needed
  • Be prepared to provide information to Qualsis about the intended use and compliance with this Policy, on Qualsis's reasonable request (Qualsis may make such requests no more than once annually unless a material incident has occurred)

Customer is responsible for all decisions made, advice given, actions taken, and failures to take action based on Customer's use of the AI Features. Qualsis does not assume responsibility for Customer's decisions, even where the decision is informed by an output of the AI Features.

4. Customer responsibility for outputs

The AI Features produce outputs that include forecasts, signals, recommendations, summaries, and other generated content (collectively, "Outputs"). Outputs are produced by machine learning models that generate probabilistic predictions based on patterns in data. Outputs may be:

  • Inaccurate, incomplete, or out of date
  • Biased in ways the Customer can or cannot anticipate
  • Reflective of training data limitations rather than ground truth
  • Inappropriate or unsuitable for a specific use the Customer has in mind
  • Subject to material change as the underlying model is updated by Qualsis or by a third-party provider

Customer is responsible for evaluating Outputs for accuracy, completeness, and appropriateness for the Customer's intended use before relying on them. Customer will not represent Outputs to third parties as more reliable than they are.

The Qualsis Services, including the AI Features, are provided as described in the Qualsis Terms of Service. The disclaimers in the Terms apply to the AI Features and Outputs, including the absence of any warranty of accuracy or fitness for any particular purpose of any AI output.

5. Compliance with AI law

Customer is responsible for complying with all laws, rules, and regulations applicable to Customer's use of the AI Features. This includes laws specific to artificial intelligence, including (without limitation):

  • The EU AI Act, in its phased application schedule
  • The UK AI regulatory framework as it develops
  • US federal AI guidance and enforcement (including the FTC's enforcement of Section 5 of the FTC Act against deceptive AI claims, and EEOC guidance on AI in hiring)
  • US state AI laws as enacted and as they enter into effect (including Colorado, Texas, and California ADMT rules, NYC Local Law 144, and others)
  • Sector-specific AI requirements applicable to the Customer's industry (financial services, healthcare, employment, education, critical infrastructure)
  • Cross-border AI requirements where applicable

Where Customer's branding, substantial modification, or change of the intended purpose of an AI system may affect either party's role under applicable AI law, Customer will reasonably cooperate with Qualsis in determining those roles and discharging the obligations applicable to each party. Nothing in this Policy reallocates an obligation that applicable law places directly on either party.

6. QoherenceAI customer-builder obligations

Upon QoherenceAI's general availability to Customer, and only then, the following obligations apply to Customer's use of QoherenceAI, unless an applicable Order or QoherenceAI-specific addendum states otherwise:

  • Governance attestation. Customer maintains documented internal governance for the agents Customer builds (specification of intended use, training data sources, evaluation methodology, incident handling, human-review checkpoints). On Qualsis's reasonable request (no more than once annually unless a material incident has occurred), Customer provides written attestation that Customer's deployed agents comply with this Policy.
  • Downstream provider compliance. Where Customer's customization or deployment of QoherenceAI takes Customer into the role of "provider" (or analogous role) under the EU AI Act or other applicable AI law, Customer is responsible for discharging the obligations of that role with respect to Customer's downstream end users.
  • Agent-level red lines. Customer's agents must honor the categorical prohibitions in Section 2 of this Policy. Customer's agents must not be configured to do, on Customer's behalf, what Customer is prohibited from doing directly.
  • Transparency to end users. Customer's agents that interact with persons or generate synthetic content must comply with applicable transparency and marking obligations, including Article 50 of the EU AI Act and any applicable transitional provisions.

These obligations may be expanded or detailed in a QoherenceAI-specific addendum to this Policy before QoherenceAI's general availability.

7. Investigation and enforcement

Qualsis may investigate suspected violations of this Policy and take action consistent with the Qualsis Terms of Service. Actions may include:

  • Requesting information about the suspected use
  • Suspending or terminating Customer's account as provided in the Terms
  • Requiring Customer to remove or modify offending content or use
  • Reporting illegal activity to law enforcement or other authorities
  • Seeking other remedies available under the Terms or applicable law

For violations that pose risk to specific persons or groups (such as deployment of an unauthorized agent that affects employment decisions, generation of harmful content targeting an identifiable person, or circumvention of safety mechanisms that produces harmful content), Qualsis may also notify the affected persons or the relevant Supervisory Authorities where required by applicable law or where reasonably necessary to prevent imminent material harm. Where legally permitted and reasonably practicable, Qualsis will notify Customer before making a notification relating to Customer Data or Customer's use of the Services.

To the extent a third-party claim arising from Customer's violation falls within Customer's indemnification obligations under the Terms, those obligations apply.

8. Qualsis's own commitments

This Policy describes obligations on Customers using the AI Features. Qualsis's own commitments about how it builds, operates, and governs AI are described in the Qualsis AI Principles at https://qualsis.com/legal/trust/ai-principles. The categorical prohibitions in Section 2 of this Policy also apply to Qualsis's own use of AI in operating the Services and the company; Qualsis does not exempt itself from rules it imposes on Customers.

Qualsis's data-handling commitments related to AI, including its restrictions on using Customer Data or Customer Personal Data to train AI or machine-learning models without the applicable explicit opt-in, are stated in the Terms and the DPA.

9. Relationship to other documents

This Policy is incorporated into the Qualsis Acceptable Use Policy by Section 4.1 of the AUP and forms part of the AUP for that purpose. Precedence among the Qualsis documents is as follows:

  • For personal-data processing matters, the DPA and its order-of-precedence provision control (and, as the DPA provides, the Standard Contractual Clauses and equivalent transfer instruments prevail over the DPA for the matters they address).
  • For all other matters, the order of precedence in Section 16.1 of the Terms applies: the Order, then the Terms, then the AUP, then other incorporated documents.
  • This Policy states in full the AI-specific prohibitions, the consequential-decision responsibility framework, and the QoherenceAI customer-builder obligations that Section 4.1 of the AUP summarizes. Consistent with Section 4.1 of the AUP (which provides that this Policy controls and contains the full statement), this Policy's full statement governs the AI-specific prohibition it describes. This Policy imposes additional restrictions on the use of AI Features; an additional restriction does not create a conflict merely because the Terms or the AUP do not restate it.
  • The AI Principles describe Qualsis's stance and commitments; this Policy describes binding rules. Where they appear to differ in substance, this Policy controls (the Principles are not intended to grant Customer rights that the binding documents do not also grant).

If a term of this Policy is held invalid or unenforceable in a particular jurisdiction, the remaining terms remain in effect; the invalid term is replaced by an enforceable term that approximates the original as closely as possible.

10. Updates

Material changes ordinarily take effect at Customer's next renewal. A material change may take effect during a current term where required by law or reasonably necessary to address a material security, safety, abuse, or regulatory risk, subject to the applicable notice, termination, and refund provisions in the Terms. The current version of this Policy is published at https://qualsis.com/legal/trust/responsible-ai, and a change log of material updates is maintained at https://qualsis.com/legal/trust/responsible-ai-changelog.

Non-material changes (clarifications, formatting, addition of examples that do not change substantive obligations) take effect upon publication.

11. Contact

For questions about this Policy, to report a violation, or to ask whether a contemplated use is permitted:

Email: Contact Privacy (for data-protection-adjacent questions); Contact Legal (for contractual questions)
Postal mail:
Qualsis, LLC
3540 Toringdon Way, Suite 200
Charlotte, NC 28277
United States

Last updated: August 28, 2026 · Effective: August 26, 2026