AI Transparency Statement

1. Who is who

Yaplet develops the AI features of the platform (the Vex AI assistant in the chat widget, AI phone agents, AI email answering, AI replies on Facebook Messenger and Instagram, and AI-generated marketing images and text) and places them on the market under its own name. For these features Yaplet is the provider in the sense of the AI Act and is responsible for designing them so that the transparency obligations are met.

A business that uses Yaplet with its own customers is the deployer. The AI Act places separate duties on the deployer, listed in section 6 below. Yaplet cannot fulfil those on your behalf, but the product is built so that meeting them is the default rather than extra work.

2. Telling people they are talking to AI (Article 50(1))

Every channel on which a Yaplet AI assistant talks to a person discloses that fact at first contact, in the language the assistant is configured to speak:

• Chat widget — the conversation header identifies the sender as an AI assistant, the empty conversation shows a plain-language notice that the visitor is chatting with an AI assistant before the first message is sent, and every AI reply carries an AI marker. When a human agent takes over, the header switches to that person's name and role.
• Phone (voice AI) — every call opens with a fixed announcement in the agent's language — in English: "This call is answered by an AI assistant." — that plays before the customer's own greeting and cannot be edited, skipped or interrupted. The assistant is also instructed to confirm that it is an AI whenever a caller asks.
• Email — AI-written email answers state that the answer was written by the company's AI assistant and, where the business allows it, offer a one-click way to reach a person.
• Facebook Messenger and Instagram — the first AI reply in a thread opens with a notice that the person is chatting with an AI assistant; the notice repeats after a break of more than 24 hours and whenever the AI takes over again after a human agent.

These notices are part of the product, not settings: the business using Yaplet cannot turn them off, and the phone announcement cannot be edited or interrupted.

3. Marking AI-generated content (Article 50(2))

Yaplet generates images and text on request (for example social media posts, newsletters and knowledge base drafts). What is marked and how:

• Images generated through Yaplet's social posting tools carry an embedded, machine-readable marker (IPTC "Digital Source Type = trained algorithmic media", stored as XMP metadata) that survives Yaplet's own resizing and logo overlay, and are declared as AI-generated to Instagram and Facebook when published through Yaplet, so those platforms show their own "AI info" label.
• AI-written text (captions, emails, articles, newsletters) is produced as a draft for a person to review and publish. There is no reliable technical standard for watermarking short text; the AI Act's exemption for text that has been reviewed by a human who takes editorial responsibility is the intended path, and the deployer duty for unreviewed public-interest text is described in section 6.
• Yaplet does not currently attach a cryptographically signed provenance manifest (such as C2PA Content Credentials) to generated images. We will update this statement if that changes.

4. What Yaplet does not do

Yaplet offers no emotion-recognition or biometric-categorisation features and does not create or alter recordings of real people (deepfakes). Nothing in the product analyses a person's face, voice or body characteristics to infer emotions or categories.

5. Code of Practice

Yaplet has not signed the European Commission's voluntary Code of Practice on AI transparency at the date of this statement. The measures above are Yaplet's own implementation of the Article 50 requirements; on request we will explain how each measure meets the requirement.

6. What stays your responsibility as the deployer

The AI Act, and the guidance of the Hungarian AI Market Surveillance Authority, place these duties on the business that uses the AI system. Yaplet supports each of them, but you must actually do them:

• Keep the AI notices in place. They are part of the product and cannot be turned off, but you must not hide them with custom styling or instruct the assistant to deny being an AI.
• Do not present the assistant as a person: do not give it a real employee's name and photo as its avatar, and do not write greetings that claim it is a human colleague.
• Make sure the assistant is configured in the language of the people it serves — the notices follow the assistant's language setting.
• If you publish AI-written text about matters of public interest (news, market or company reports, public information) without a human reviewing it and taking editorial responsibility, label it as AI-generated.
• If you publish images, audio or video that depict real people, places or events in a way that looks authentic but was created or altered by AI, label them clearly as AI-generated — regardless of which tool produced them.
• Keep this statement, dated, in your compliance records as the provider's description of the built-in transparency measures. Contractual terms are in the Yaplet Terms of Service ("AI and Automated Features").

7. Written confirmation

This page is Yaplet's written statement, for the benefit of its customers, that the AI features of the Yaplet platform are designed to meet the transparency obligations of Article 50 of Regulation (EU) 2024/1689 as described above. It is versioned and dated; earlier versions are available on request.

Questions from customers, auditors or authorities: support@yaplet.com

8. References

• Regulation (EU) 2024/1689 (the AI Act) — official text
• Hungarian AI Market Surveillance Authority — transparency guidance (mihivatal.gov.hu)
• Yaplet Terms of Service — AI and Automated Features