AI Usage Policy

1. Purpose

This AI Usage Policy explains how BECKO - TRAVEL OS LTD ("Becko") deploys artificial-intelligence features inside the Service, how customer data is handled in those workflows, and the responsibilities of users.

2. AI Capabilities

The platform uses AI for, among other things:

• Automated responses in Messenger, Instagram, and WhatsApp conversations
• AI agents that assist with flight, hotel, and travel package recommendations
• Quotation, itinerary, and email drafting
• Creative generation in Becko Studio (images, video scenes, captions, templates)
• Lead enrichment, conversation summarisation, and quality scoring

3. Models & Providers

• OpenAI (gpt-4o-mini and equivalents) — chat, agents, summarisation, drafting
• Google Gemini (via Lovable AI Gateway) — Becko Studio, travel-card generation, multimodal tasks
• DALL-E 3 — image generation
• Additional models may be added; we will reflect them in our subprocessor list

API providers we use contractually agree not to train their foundation models on customer API data. Becko does not use customer conversations to train public AI foundation models.

4. PII Anonymization

Any internal AI improvement, pattern extraction, or quality-review workflow strips personally identifiable information before processing. Identifiers such as names, email addresses, phone numbers, and payment card numbers are replaced with anonymous placeholders (e.g. [CUSTOMER], [EMAIL], [PHONE], [CARD]).

• Anonymized data cannot be traced back to individuals
• Raw conversation data containing PII is never used for model training or fine-tuning
• Each anonymization event is logged with metadata (number of tokens stripped) for audit purposes

5. Limitations & Hallucinations

AI features are probabilistic and may produce inaccurate, incomplete, biased, or misleading outputs ("hallucinations"). The Service should not be relied upon as the sole source of truth for:

• Prices, availability, fees, taxes, or refund amounts
• Flight schedules, baggage rules, visa requirements, or health regulations
• Legal, medical, financial, or tax advice
• Customer commitments, contracts, or binding offers

Users are solely responsible for reviewing and verifying AI-generated outputs before transmitting them to customers, suppliers, or third parties.

6. Human Oversight

• Brand administrators control whether AI agents are enabled per channel
• AI agents can hand off to human staff at any point
• Customers can opt out of AI-assisted conversations by replying with STOP or by requesting a human agent
• Significant business actions (refunds, cancellations, contract changes) should require human approval

7. Prohibited Uses

You must not use Becko AI features to:

• Deceive users about whether they are talking to a human
• Generate harmful, discriminatory, sexual, violent, or unlawful content
• Generate content that infringes intellectual property or impersonates real persons
• Make automated decisions with legal or similarly significant effects on individuals without appropriate safeguards under Article 22 GDPR
• Bypass safety controls, prompt-inject, or extract model weights / training data

See also our Acceptable Use Policy §4.

8. Monitoring & Moderation

We may apply automated moderation to AI inputs and outputs to detect abuse, prompt injection, malware, illegal content, or policy violations. Such monitoring is performed on aggregated, anonymized, or pseudonymised data wherever feasible.

9. EU AI Act & Regulatory Compliance

Becko monitors emerging AI regulations including the EU AI Act, the UK AI regulatory framework, and relevant US state laws. We classify our AI features and apply appropriate transparency, documentation, and risk-management measures consistent with the regulatory category that applies.

10. Contact

Questions, opt-out requests, or AI-related complaints: privacy@becko.app.