ZipZy® AI Usage Policy

1.1 Scope. This Policy applies to every AI-assisted capability across the Services, whether currently offered or released in future — including generative writing assistance, summarisation, translation, content suggestions, image generation, classification and tagging, analytics interpretation, message drafting, form and page generation, and conversational assistants.

1.2 Definitions

2.1 Inputs. You retain all rights in your Inputs. Inputs are Customer Data under the Terms of Service.

2.2 Outputs. As between you and us, and subject to third-party rights and applicable law, we assign to you all right, title, and interest we may hold in Outputs generated for you. You may use Outputs for any lawful purpose consistent with this Policy.

2.3 Non-Uniqueness. Generative models may produce identical or substantially similar Outputs for different users submitting similar Inputs. We make no representation that any Output is unique, novel, or exclusive to you, and we cannot assign rights in Outputs also generated for others.

2.4 Copyrightability. The legal status of AI-generated material varies by jurisdiction and may not attract copyright protection. You are responsible for assessing this where it matters to your use.

2.5 Our Use of Outputs. We do not claim ownership of Outputs and do not use them for our own commercial purposes beyond providing and securing the Services.

3.1 Probabilistic by Nature. AI Features generate statistically likely content. They do not verify facts, consult authoritative sources at generation time, or possess understanding in the human sense.

3.2 Known Limitations. Outputs may be factually incorrect, fabricated but plausible (“hallucinated”), outdated relative to real-world events, internally inconsistent, biased in ways reflecting training data, incomplete, inappropriate for a specialised context, or non-compliant with regulations applicable to you.

3.3 Not Professional Advice. Outputs are not legal, medical, financial, tax, accounting, employment law, immigration, insurance, safety, or other professional advice, and must not be presented as such to any third party.

3.4 Verification Required. Verify factual claims, figures, calculations, citations, legal or regulatory statements, contact details, and code before relying on or publishing an Output.

3.5 No Warranty. AI Features are provided without warranty as to accuracy, completeness, reliability, fitness for purpose, or non-infringement. The disclaimers and liability limitations in the Terms of Service apply in full.

You are solely responsible for:

• the Inputs you submit, including any personal data, confidential information, or third-party content contained in them, and for having the right to submit them;
• reviewing and verifying Outputs before use;
• any content you publish, distribute, or act upon that was generated with AI assistance;
• compliance with laws applicable to your use — including advertising standards, consumer protection, financial promotion rules, medical and pharmaceutical advertising restrictions, employment law, electoral law, and data protection;
• any harm arising from reliance on an unverified Output; and
• ensuring your own users and employees understand that AI Features are used, where that matters to them.

4.1 Confidential Information. Consider carefully before submitting confidential, privileged, trade secret, or export-controlled information as an Input. Do not submit payment card data, government identifiers, health records, or biometric data unless the specific module is designed to process them.

5.1 Mandatory Human Review. You must not use AI Features to make, or as the sole determining factor in making, decisions producing legal or similarly significant effects on an individual. This includes decisions on hiring, promotion, discipline, termination, compensation, performance rating, credit, insurance, housing, education, healthcare, or access to essential services.

5.2 Meaningful Review. Where an AI Feature assists such a decision, a qualified human must have the authority, competence, information, and time to reach a different conclusion, and must actually consider whether to do so. Rubber-stamping an Output is not meaningful review.

5.3 HRMS Module. Where AI Features are used within HRMS functionality, you remain the employer and the Data Fiduciary. You are responsible for compliance with employment law, non-discrimination obligations, and any duty to inform or consult employees about automated tooling.

5.4 Records. We recommend you retain records of human review for decisions affecting individuals, in case you are required to demonstrate compliance.

You must not use AI Features to generate, refine, translate, or distribute:

6.1 Deception and Manipulation

• Disinformation, fabricated news, or synthetic media presented as authentic
• Content impersonating a real individual, including their voice, likeness, writing style, or signature, without their documented consent
• Content impersonating an organisation, government body, or ZipZy® itself
• Fake reviews, testimonials, endorsements, engagement, or social proof
• Phishing messages, pretexting scripts, fraudulent invoices, or social engineering content
• Content designed to manipulate elections, suppress voting, or mislead about electoral processes

6.2 Harm and Illegality

• Malware, exploits, ransomware, obfuscation of malicious code, or attack tooling
• Instructions for weapons, explosives, or dangerous chemical, biological, radiological, or nuclear agents
• Content facilitating human trafficking, exploitation, or terrorism
• Content that sexualises minors in any form
• Non-consensual intimate imagery or sexualised depictions of real people
• Content promoting self-harm, suicide, or disordered eating
• Harassment, threats, doxxing, or targeted abuse

6.3 Discrimination

• Content or automated evaluation that discriminates unlawfully on the basis of caste, religion, race, sex, gender, sexual orientation, disability, age, place of birth, or any other protected characteristic
• Scoring, ranking, or profiling of individuals for social control purposes

6.4 Surveillance and Privacy

• Biometric identification or emotion inference about individuals without a lawful basis and consent
• Analysis intended to infer sensitive attributes — health status, sexual orientation, religious or political belief — about identifiable individuals
• Covert monitoring of employees or individuals beyond what is lawful and disclosed

6.5 Professional Misrepresentation

• Legal, medical, financial, or other professional advice presented to third parties as authoritative without qualified human review and appropriate disclaimers
• Regulatory filings, contracts, prescriptions, or diagnoses issued without qualified review

6.6 Technical Misuse

• Attempts to extract system prompts, model weights, or training data
• Prompt injection, jailbreaking, or circumventing safety controls
• Using AI Features to train, fine-tune, distil, or benchmark a competing model
• Automated bulk generation to overwhelm capacity or evade usage limits

6.7 Academic and Assessment Integrity

• Content intended to be submitted as a person's own original work in violation of an institution's rules

7.1 No Training on Your Data. We do not use Customer Content, Inputs, or Outputs to train, fine-tune, or improve foundation models.

7.2 Subprocessor Obligations. We contractually require every Model Provider processing data on our behalf to refrain from training on that data and to delete it after processing, subject only to short retention for abuse monitoring where required by that provider's own safety obligations. Providers unwilling to accept these terms are not engaged.

7.3 Aggregate Metrics. We use aggregated, non-identifying operational metrics — feature adoption, latency, error rates, acceptance and rejection rates — to improve the product. These metrics do not include the substance of your Inputs or Outputs.

7.4 Explicit Opt-In. If we ever offer a programme in which Customers may voluntarily contribute data to improve a feature, participation will be strictly opt-in, separately described, and revocable at any time. It will never be the default.

7.5 Feedback. If you submit an Output as feedback — for example, by flagging it as inaccurate or unsafe — we may review that specific item to improve safety controls and product quality. Feedback submission is a deliberate action you take.

8.1 Processing. Inputs are transmitted to the relevant model, processed to generate an Output, and returned to you.

8.2 Retention. Inputs and Outputs are not retained beyond what is necessary. Limited copies may be held briefly for abuse detection, safety review, and debugging and are then deleted. Where you save an Output into your workspace, it becomes Customer Content under your control and is retained per the Data Retention Policy.

8.3 Encryption. Inputs and Outputs are encrypted in transit using TLS 1.2 or higher and at rest where stored.

8.4 Access. Our personnel do not read Inputs or Outputs except where necessary to investigate a support request you raise, to investigate suspected abuse, or where required by law. Such access is logged.

8.5 Processor Role. Where AI Features process personal data on your behalf, we act as Data Processor under the Data Processing Addendum. You are responsible for the lawful basis.

8.6 Cross-Border. Some Model Providers operate outside India. Transfers comply with Section 16 of the DPDP Act and Section 10 of the Privacy Policy.

We apply content safety filters to Inputs and Outputs, rate limits and anomaly detection to prevent bulk misuse, prompt-injection defences on features that process untrusted content, and logging of safety events. Where an Input or Output is blocked, you may receive a generic refusal without detailed explanation, to avoid assisting circumvention. Repeated triggering of safety controls is reviewed and may result in restriction of AI Features on the Account.

Safety controls are imperfect. They may occasionally block benign requests or fail to block harmful ones. They do not transfer responsibility for compliance from you to us.

AI Features may be delivered using models operated by third parties, listed in the Subprocessor List Policy. We select providers on the basis of capability, security posture, data handling commitments, and reliability, and we may substitute or add providers. Availability of any specific model, version, or capability is not guaranteed, and model behaviour may change between versions in ways that affect Output style or quality.

11.1 Third-Party Rights. Outputs may resemble existing works. You are responsible for ensuring that your use of an Output does not infringe copyright, trademark, design, publicity, or other rights.

11.2 Trademarks and Likeness. Do not use AI Features to generate content incorporating third-party trademarks, brand assets, or the likeness of real people without authorisation.

11.3 Your Inputs. By submitting an Input you represent that you have the rights necessary to do so, including for any third-party content it contains.

11.4 Complaints. Copyright complaints relating to AI-assisted content are handled under the Copyright Policy; trademark complaints under the Trademark Policy.

12.1 In-Product Labelling. AI-assisted functionality is identified in the interface at the point of use.

12.2 Your Disclosure Obligations. You are responsible for any disclosure obligations that apply to you — for example, informing recipients that a message was AI-generated where required by law, platform rules, or professional standards. We recommend disclosure wherever an audience would reasonably expect human authorship.

12.3 Documentation. Capabilities, limitations, and applicable model families for each AI Feature are described in the Documentation.

AI Features are subject to plan-based usage entitlements, expressed as credits, tokens, requests, or generations. Usage is visible in your Account. Exceeding an entitlement may result in temporary unavailability of AI Features until the next cycle, or overage billing where disclosed. Bulk or automated invocation to circumvent limits violates this Policy and the Acceptable Use Policy.

AI Features depend on third-party infrastructure and may be degraded, rate limited, or temporarily unavailable independently of the rest of the Services. Unless expressly stated otherwise, AI Features are excluded from the availability commitments in the Service Level Agreement. Features designated beta or preview may change or be withdrawn at any time.

Violations of this Policy are enforced under Section 9 of the Acceptable Use Policy, and may result in blocking of specific requests, restriction or removal of AI Features from the Account, suspension, or permanent termination without refund. Violations involving child safety or illegal content are reported to law enforcement.

Appeals may be sent to abuse@zipzy.in.

AI capability and regulation are evolving quickly. We may update this Policy as models, features, and legal requirements change. Material changes take effect 30 days after posting, except changes required to address safety risks or comply with law, which take effect immediately.

Pranava Company Group (ZipZy®)
62, Mandola, Loni, Ghaziabad, Uttar Pradesh 201102, Bharat (India)