Synthetic Media Policy

Last updated: 2026-02-22

1. What This Policy Covers

AfterLight generates synthetic media as part of its reflection experience. This includes AI-generated text responses, abstract visual imagery, voice narration, and photo-based micro-animations. This policy explains what we generate, how we safeguard against misuse, and what your rights are.

2. Types of Synthetic Media

AI-Generated Text

Ben, AfterLight’s AI reflection companion, generates text responses based on the memories you share. These responses are clearly labeled as “Ben (AI reflection)” in the interface. Ben is powered by OpenAI and operates within strict ethical guardrails.

Abstract Visual Imagery

When creating embodied reflections, AfterLight may generate abstract atmospheric images inspired by your memory descriptions. These images never depict real people — the generation system explicitly blocks person generation and uses safety filters to ensure abstract, non-representational output.

Voice Narration

AfterLight can generate voice narrations of reflection text using Google Cloud Text-to-Speech. The voice is a standard AI voice (not cloned from any person) and is clearly presented as AI-generated audio.

Photo-Based Micro-Animation

With your explicit consent, AfterLight can create subtle animated sequences from a photo you upload. These animations add gentle motion (such as a blink or slight head movement) to a still image. This feature uses neural animation technology and requires your consent before processing.

3. Ethical Safeguards

All synthetic media generation in AfterLight is governed by strict safeguards:

• All AI-generated content is clearly disclosed and labeled as such • Image generation blocks depictions of real people (person_generation=dont_allow) • Safety filters are set to the highest level (block_most) • Voice is a standard AI voice — no voice cloning or imitation of any person • Photo animation requires explicit consent before processing • Photo animation is limited to the photo you provide — no face swapping or cross-face generation • All animated output includes an ethical disclosure label

4. Rate Limits and Cost Controls

To prevent misuse and manage resources responsibly:

• Micro-animations are limited to 3 per day per user • Embodiment creation is limited to 10 per day per user • Neural animation processing has a daily cost budget • All processing requires authentication

5. What We Do Not Generate

AfterLight will never:

• Clone or imitate the voice of a real person • Generate realistic images of real people • Create deepfakes or misleading synthetic media • Generate sexual, violent, or inappropriate synthetic content • Process biometric data for identification purposes • Generate synthetic media without user initiation and consent

6. Your Rights

• You can delete any synthetic media (embodiments, animations) at any time from the app • You can export all your data, including metadata about generated content, via Settings • You can delete your account and all associated content at any time • You can choose not to use any synthetic media features — they are entirely optional

7. EU AI Act Compliance

Under the EU AI Act, synthetic media generation is subject to transparency obligations (Article 50). AfterLight complies by clearly labeling all AI-generated content, providing this dedicated disclosure, and ensuring users are aware before any synthetic content is created.

8. Contact

Questions about our synthetic media practices: ethics@umbrella-research.org