AI Disclosure

Document control

1.Who these terms are with — the Customer

1.1 The App is contracted, operated, and paid for by the Customer: the account holder, who must be an adult — at least 18 years old, or the age of majority in the place where the Customer resides, whichever is greater. Registering for or using the App confirms that the Customer meets this requirement. The App is not offered to, and may not be operated by, a person under that age.

1.2 Throughout this AI Disclosure, "you" means the Customer. A song, lyric, or cover image may be personalized for any person whose first name the Customer provides — a child, another member of the household, or the Customer themselves. The person for whom content is created is a beneficiary of that content and is not a party to these terms and not an operator of the App.

1.3 Where the content is for, or references, a child. Content is often created for children, and a child may hear a finished song when the adult Customer plays it for them; a child does not, however, operate the App, hold an account, or provide data to us. Because content is made for children, we address child protection proportionately to this adult‑operated model: we do not market or advertise to, track, or profile children; we keep the data about any child a song is for to a first name (Section 4); and we apply child‑safety filtering to every generation (see Sections 6 and 8). Where the Customer provides the first name of, or otherwise refers to, a minor, the Customer represents that they are that minor's parent or legal guardian and consents, on the child's behalf, to the processing described in this AI Disclosure and in our Privacy Notice. This is a lawful, parent‑operated processing model; it does not remove any child‑protection obligation, and the child‑safety, consent, and data‑minimization measures in this document apply in full.

1.4 Incidental, parent‑supervised access does not make the App child‑directed. As with any general‑audience app, an adult Customer may let a child hear a song on the adult's own device; that incidental, supervised listening is not a child using or accessing the App as a user, and it does not convert this adult‑operated App into a service directed to children, or give us actual knowledge that a child is a user, under the Children's Online Privacy Protection Act. Three affirmative facts about how the App is built hold this line: (1) we build no profile of the child — we hold only the first name the adult provides (Section 4); (2) we direct no feature, screen, content, character, or message at a child — there is no child login, no child‑facing mode, and nothing that invites a child to act, earn, or transact; and (3) we collect no data from the child — every input is provided by, and the account is operated by, the adult Customer. Honest boundary, which we keep and do not over‑claim: a minor's first name is processed, with the adult Customer's parental consent (Section 1.3), and we still do not market or advertise to, track, or profile a child. If a future feature were to speak to a child, or be operated by a child, that would change this analysis, and we would re‑assess it before that feature ships.

2.How content is created with AI

2.1 Content created through the App is generated by artificial intelligence. It is not recorded by human performers and is not drawn by a human illustrator:

• The song audio is composed and synthesized by an AI music model. No human musician performs on the track; the singing voice is synthesized by the model and is not recorded from any person.
• The lyrics are written by an AI language model from the theme, dedication, first name, and other inputs provided in the creation flow.
• The cover image is generated by an AI image model.

2.2 AI systems assemble new content statistically from patterns learned during training. Output is therefore novel and probabilistic: identical inputs can produce different results, output can contain inaccuracies or unexpected elements, and quality varies. AI‑generated content is a creative, entertainment work and must not be relied on as factual, educational, medical, therapeutic, legal, or professional advice (see Section 12).

3.AI providers and models we use

3.1 We use the following third‑party AI providers, each located in the United States. This table is descriptive; the authoritative, current list is our Subprocessor List, published with and linked from our Privacy Notice, which is the source of truth. We may add, change, or remove providers as the technology evolves (see Section 16); we do so by updating the Subprocessor List and giving notice of material changes, not by re‑drafting this document or re‑collecting consent, because the consent described in Section 5 names providers by reference to that list.

3.2 Honest statement of the output‑IP risk (music). Our providers license the content their models generate for our commercial use, provide content‑authenticity tooling (including the SynthID watermark on the audio), and apply safety filters, which we keep active. However, the music provider does not indemnify Happy Songs for the intellectual‑property content of the music output. AI music generated from large training datasets carries a residual risk that it may resemble or incorporate third‑party protected material — a known, industry‑wide issue — and that risk is borne by Happy Songs, not by the AI provider. We reduce and bound this risk (licensed output, safety filters kept on, prohibition and screening against imitation of real artists under Section 11, provenance/watermarking under Section 6, a notice‑and‑takedown process under Section 13, and the media‑liability / IP‑infringement insurance we maintain or intend to maintain as a financial backstop). We do not and cannot eliminate it, and we make no warranty that any output is original, unique, exclusive, or non‑infringing (see Sections 9 and 13, and the AS‑IS and limitation‑of‑liability provisions of the Terms).

3.3 Data‑breach risk. No AI provider indemnifies Happy Songs against a personal‑data breach. Happy Songs therefore carries residual data‑security risk, which it mitigates through its information‑security program, its data‑processing agreements with each provider, and the cyber / breach‑response insurance it maintains or intends to maintain. This is a separate risk from the output‑IP risk in Section 3.2 and is not covered by any provider's copyright commitment.

4.What we hold, and what we send to AI providers — data minimization

4.1 About the person a song is for, we store only a first name. We do not store a surname, a nickname, an age, a date of birth, a profile, a photograph, a voice recording, or precise location for that person. Any age band or life stage you select is used only in the moment to help you browse occasions and tailor the song; it is not stored as a data point and is not used to build a profile.

4.2 From the Customer, we hold: the phone number used to log in (the primary account identifier); an email address (optional); device and technical data; usage and analytics data (consent‑gated, off by default, and never used to track or profile the child a song is for); the content you submit to create a song; and your communications with us. These are described in full in our Privacy Notice.

4.3 No special‑category data of any person, of any age. The App does not ask for and does not knowingly process health, medical, biometric, genetic, religious, sexual‑orientation, or other special‑category / sensitive data of the person a song is for or of the Customer. In particular, the "Get Well" / "Mejórate Pronto" occasion captures only a generic well‑wishing sentiment; it never asks for, records, or infers a specific illness, diagnosis, symptom, or health condition, and it is not a health service.

4.4 What is sent to the AI providers, and the placeholder‑name minimization. When you create content, the App sends only what is needed to generate it, and it minimizes the personal data each provider receives:

• To the music provider (Google Lyria / Vertex AI) — the creation prompt, which includes the real first name (because the name is sung in the song) together with the theme, dedication, and lyric content — for the purpose of generating the song audio.
• To the lyric providers (Anthropic and OpenAI) — the theme, dedication, and lyric prompt, in which the real first name is replaced by a placeholder name; the real first name is inserted locally, in the App, after generation, so that the real name of the person a song is for is not sent to the lyric providers.
• To the cover‑art provider — an image prompt derived from your inputs, which does not include the real first name where feasible; the real first name is applied locally.

4.5 The App does not send the Customer's phone number or other direct account identifiers to the AI providers as part of the creation prompt.

4.6 No training on your inputs. Under our agreements with these providers (all contracted on their enterprise / API tier), they do not use the content you submit through the App to train their own general‑purpose AI models. Providers may retain content briefly to operate the service and to detect abuse and enforce their safety policies, as set out in their terms and our data‑processing agreements; content flagged by a provider's safety systems may be retained for a longer, defined period under that provider's policy. Retention of your data by Happy Songs is governed by our Privacy Notice and Data Retention Policy.

5.Consent before your data is sent to AI providers (Apple App Store Guideline 5.1.2(i))

5.1 Because the App sends information you enter to third‑party AI providers, the App obtains your explicit, affirmative, in‑app consent before it does so, and it names the providers at the point of consent. A link to a policy, by itself, is not sufficient consent for this purpose.

5.2 The consent screen states what is sent and to whom (as described in Section 4.4), names the current AI providers by reference to the Subprocessor List, and confirms the minimization measures. You may decline. If you decline, the App cannot generate content, because generation requires sending your inputs to these providers; declining does not otherwise limit your use of the App's non‑AI features. This consent is separate from, and additional to, the consents described in our Privacy Notice and any parental‑consent flow.

5.3 Please keep prompts free of unnecessary personal or sensitive information. You do not need to enter sensitive data to create a song — a first name, an occasion, and a theme are enough. Do not enter a home address, a phone number, health details, financial information, a government identifier, or the special‑category data of any person into a prompt.

6.Content authenticity — the SynthID watermark and AI labeling

6.1 Audio watermark (SynthID). Song audio generated through the App carries Google's SynthID watermark — an embedded, machine‑readable marker that identifies the audio as AI‑generated. The watermark is designed to be robust and to persist through common transformations, so the audio remains identifiable as AI‑generated even if a copy were extracted from the App — which the Terms prohibit (see Sections 9 and 10), since songs are provided by in‑App access only and are not offered as a downloadable file.

6.2 In‑App labeling. The App labels AI‑generated content as AI‑generated in the interface, including at the point of creation and during playback, so that the disclosure travels with the content and is present where you experience it.

6.3 Public identification. In addition to the SynthID watermark and the in‑App label, Happy Songs may publicly and prominently identify content created through the App as AI‑generated, and may apply human‑readable AI labels and provenance credentials, wherever such content is surfaced, referenced, or shared by us.

6.4 Do not remove or defeat the watermark or label. You agree not to remove, alter, obscure, or attempt to defeat the SynthID watermark, any provenance metadata, or any AI‑generated label.

7.Rejected requests, free promotional access, and paid generations

7.1 Happy Songs is a paid subscription service. It is currently offered with free promotional access: during the promotional period no payment card is collected and no charge is applied. Where paid plans are offered, the Customer is the payer and the account holder.

7.2 Every creation request is screened by automated safety and content filters (see Section 8). Because AI systems and safety filters are probabilistic, we cannot and do not guarantee that every request will produce a finished song, lyric set, or cover image, or that the filters will catch every instance of unwanted content.

7.3 If a request is rejected:

• During free promotional access, no payment card is collected and no charge is applied, so a rejected request results in no charge.
• Where a paid generation is offered and a request is rejected by the safety filters, you are entitled to a refund or a retry in accordance with the Terms, and we disclose this before you pay. We do not knowingly charge for a paid generation that is not delivered.

8.Content moderation, child‑safety filters, and automated decisions

8.1 Every creation request runs through automated child‑safety and content filters (including the safety filters of the music provider and our own moderation layer). These filters are the first line of protection for a product that children are likely to access, and we keep them active at all times.

8.2 A rejection by these filters is an automated decision made, at the moment of generation, without human involvement. The basic logic is that your inputs are screened against safety and content rules, and a request is blocked if it matches a prohibited category — for example, content that is sexual, violent, hateful, harassing, deceptive, or that attempts to imitate a real, identifiable person (see Section 11).

8.3 Human review. If a request is rejected and you believe the decision was in error, you may request human review and contest the outcome by contacting support@happysongs.ai. We provide this on request; where the law of a market in scope requires human review of an automated decision, we provide it as required.

8.4 If content that should have been blocked reaches you, or you encounter content you believe is inappropriate, use the in‑App report / flag control or contact support@happysongs.ai. See also our Acceptable Use Policy and Child Safety / Trust & Safety Policy.

9.Rights in AI‑generated content — a revocable access license, not ownership

9.1 No exclusive copyright. A work generated entirely by AI, without sufficient human authorship, generally cannot be registered for copyright and is not owned exclusively by us or by you. We therefore cannot grant, and do not grant, an exclusive copyright in any song, lyric, or cover image.

9.2 What you receive. You receive a personal, non‑exclusive, non‑transferable, revocable license to access and play the content you create, for personal and non‑commercial purposes, only inside the App, as set out in the Terms. You are not given ownership, an exclusive right, or a downloadable, exportable file.

9.3 Closed ecosystem; offline cache. For your convenience, the App may cache your songs on your device for offline playback inside the App only (App‑managed). You may not export, download as a file, copy, re‑host, share, or otherwise remove the content from the App. When your access ends — on cancellation, on account deletion, or on revocation — the App stops playback and purges the on‑device cache, and the server copy is deleted in accordance with our Data Retention Policy.

9.4 Meaning of "yours" / "forever." Where our copy says a song is "yours," "yours forever," or "for always," it means your continued in‑App access while your access is active — not exclusive title, and not a file you keep outside the App. Access is not perpetual: on a paid plan it continues while the subscription is active; during free promotional access it is granted at our discretion (with no payment card collected and no auto‑charge); and it may be revoked, all as set out in the Terms. Rights in content created during free promotional access do not retroactively expand when paid plans launch.

9.5 Not unique, not one‑of‑a‑kind. Some songs are name‑personalized templates: an existing track is personalized with the first name you provide, and the same underlying song is reused and shared across many Customers — including other Customers whose child shares the same first name, who may receive the same song. We therefore do not describe any song as "unique," "one‑of‑a‑kind," or "exclusive." Your song is personalized for you with the first name you provide; it is not exclusive to you. Any marketing or in‑App language suggesting a song is "yours," "exclusive," or "owned" is to be read consistently with this Section and the Terms.

10.Anti‑circumvention — extracting content outside the App

10.1 Access to the songs is controlled by technical measures within the App. You must not circumvent, disable, bypass, or attempt to defeat those measures, and you must not extract, capture, copy, re‑host, redistribute, or share any song, lyric, or cover image outside the App.

10.2 Doing so is (a) a breach of these terms and the Terms; (b) an unauthorized use of the App and our systems; and (c) may violate the anti‑circumvention provisions of 17 U.S.C. § 1201 (DMCA) and analogous laws, to the extent an access‑control technical protection measure is in place. This anti‑circumvention protection is distinct from copyright in the underlying work; because a fully AI‑generated work is generally not registrable, we do not frame extraction as copyright infringement of a Happy Songs work.

11.Prohibited uses — no imitation of real people

11.1 You may not use the App to create content that:

• imitates, clones, or reproduces the voice, name, likeness, image, or distinctive performing style of a real, identifiable person, including any celebrity, recording artist, musician, or public figure;
• names or references a real artist or celebrity in order to mimic their style or voice;
• infringes any third party's copyright, trademark, right of publicity, moral rights, or privacy rights; or
• is sexual, involves the sexualization of minors, or is violent, hateful, harassing, defamatory, deceptive, or otherwise prohibited under our Acceptable Use Policy.

11.2 No real artists or brands in any input. Do not enter the name of a real artist, band, song title, celebrity, or brand, or an "in the style of a real artist" instruction, anywhere in a prompt, dedication, or theme. Genre and mood are fine ("an upbeat pop‑style song," "a gentle lullaby"); a real person's or brand's identity is not.

11.3 Name‑personalization is screened. Because some songs are personalized with a first name you provide (Section 9.5), and because a first name can also be a celebrity or brand name, our name catalog and name‑entry path are screened so that a name is treated as a personal first name, not as a reference to a public figure or brand.

11.4 We apply filters (including a real‑name / celebrity‑name filter) to block prohibited requests, and we operate a notice‑and‑takedown process (Section 13). Content that violates this Section may be blocked or removed and may result in suspension or termination under the Terms. These restrictions reflect, among others, right‑of‑publicity and anti‑impersonation laws in the United States and Mexico. You are responsible for your inputs and, as set out in the Terms, you indemnify Happy Songs against claims arising from the content you direct the App to create — an indemnity limited to your own inputs and misuse, which does not transfer to you the output‑IP risk described in Sections 3.2 and 13.

12.Accuracy, limitations, and appropriate use

12.1 AI output can be wrong. Lyrics, names, facts, and images may be inaccurate, unexpected, or nonsensical. Review content before sharing it with, or playing it for, a child.

12.2 Not professional advice. Content is entertainment. It is not medical, psychological, therapeutic, educational‑assessment, legal, or financial advice, and must not be relied on as such. Any theme suggesting well‑being or "feel better" content is not a health service (see Section 4.3).

12.3 Safety and remediation. Because content is intended for an audience that includes children, we maintain safety filtering and a remediation channel (report / flag, takedown, and support). If AI content causes concern, stop using it and contact support@happysongs.ai. This is provided in addition to, and does not waive, any non‑excludable statutory rights you may have.

13.Training data, third‑party rights, and our risk posture

13.1 AI models are trained on large datasets, and there is active industry litigation about training data and generated‑music similarity. Being straight with you: for the music, the AI provider does not indemnify us against third‑party IP claims, so Happy Songs carries that risk. To manage it we keep provider safety and content filters active; prohibit and technically screen against imitation of real artists, voices, and styles (Section 11); operate a notice‑and‑takedown process; preserve SynthID and provenance signals (Section 6); and maintain or intend to maintain media‑liability / IP‑infringement insurance as a financial backstop. These measures reduce and bound the risk; they do not eliminate it. We do not warrant that AI output is original, unique, exclusive, or free of all third‑party rights.

13.2 The content the App provides is offered AS IS and AS AVAILABLE, without warranty of any kind, and Happy Songs' liability is limited as set out in the Terms. Nothing in this AI Disclosure transfers the output‑IP risk to you; your indemnity under Section 11 and the Terms is limited to your own inputs and misuse.

13.3 If you believe content infringes your rights, contact our designated IP agent at privacy@happysongs.ai and see our Acceptable Use Policy (takedown).

13.4 Voice and biometric data — affirmative commitment. The App does not capture, request, or process anyone's voice or voiceprint; the singing voice on every song is synthesized by AI, not recorded from you or from any child. As a commitment that stands regardless of future features, we will not use any audio to build, train, or improve voice‑cloning or voice‑replication technology, and we will not repurpose content created through the App to clone a real person's voice. If we ever introduce a feature that captures voice, we will first obtain separate, explicit consent and publish the applicable biometric‑data terms.

14.Provenance record

For each song created through the App, we record which AI provider and model generated it and retain the associated SynthID / provenance signal. We keep this record to support content authenticity, risk allocation and liability tracking, safety, and takedown processes, in accordance with our Privacy Notice and Data Retention Policy.

15.Payments and access

15.1 Happy Songs is a paid subscription service, currently offered with free promotional access (Section 7.1). Happy Songs does not store card numbers. Where subscriptions are purchased through the app stores, Apple and Google process the card through in‑app purchase, and we receive only subscription status and receipts (via our subscription‑management provider). The App has no separate web or card payment processor; all subscription charges are processed by the app stores through in‑app purchase.

15.2 We store transaction and subscription records and the tax records the law requires us to keep. Payment data handling is described in our Privacy Notice.

16.Changes to the service, providers, and this AI Disclosure

16.1 We may add, change, or discontinue AI providers, models, features, plans, and pricing at any time. Any such change applies prospectively and does not retroactively reduce the rights in, or access to, content you have already created and paid for, except as permitted by these terms. Where the law requires notice — including, in particular, notice of a subscription price increase — we give it, and you may cancel before the change takes effect.

16.2 Changes to our AI providers are made by updating the Subprocessor List (the source of truth) and giving notice of material changes, without re‑drafting this document and without re‑collecting the consent described in Section 5, because that consent names providers by reference to the Subprocessor List.

16.3 We may update this AI Disclosure as our providers, models, features, or legal obligations change. Material changes are notified through the App and/or by other reasonable means, and the "Effective date" above is updated. Continued use after an update constitutes acceptance to the extent permitted by law and the Terms.

17.Governing terms, jurisdiction, and contact

17.1 Markets in scope. This AI Disclosure covers the App as offered in the United States and Mexico. In the United States, our processing is subject to, among others, the Children's Online Privacy Protection Act and the amended COPPA Rule, the Texas Data Privacy and Security Act (which has no small‑business threshold), and the California Consumer Privacy Act / CPRA and comparable state laws. In Mexico, our processing is subject to the Ley Federal de Protección de Datos Personales en Posesión de los Particulares (published in the Diario Oficial de la Federación on 20 March 2025), and the competent authority is the Secretaría de Anticorrupción y Buen Gobierno. A Spanish (Mexico) version of this AI Disclosure is published for the Mexican market; in case of discrepancy, the versions are to be read as equivalent in meaning. Where the App is offered in other markets, additional market‑specific terms apply and are published separately.

17.2 Contact.

Happy Songs USA Corp.
Registered address: 8350 Ashlane Way, Suite 103, The Woodlands, TX 77382, United States.
Operating address: Calle Tijuana 22‑1, Col. Del Valle, C.P. 03100, Benito Juárez, Ciudad de México, México.
Website: happysongs.ai