Ethical AI in Memorial Technology: Our Approach

As AI technology advances, the possibility of creating digital representations of loved ones is no longer science fiction. But with this capability comes profound responsibility. At LifeDraft, we've spent considerable time thinking about how to build memorial AI technology that respects human dignity, maintains clear boundaries, and genuinely serves the needs of grieving families.

This post outlines our ethical framework and the principles that guide every technical and design decision we make.

The Core Question: What Should Memorial AI Be?

Before building anything, we asked ourselves a fundamental question: What should a memorial AI representation actually be?

The answer, we believe, is not to recreate or replace a person. No AI can capture the full essence of a human being—their consciousness, their agency, their continued growth and change. Attempting to create such a facsimile would be both technically impossible and ethically problematic.

Instead, we believe memorial AI should be a thoughtfully curated collection of someone's own words, stories, and voice—presented honestly as a memorial representation, not as the person themselves.

"A memorial representation should be like a beautifully crafted book of memories—authentic, curated by its subject, and always clearly a tribute rather than a resurrection."

Principle 1: Your Words Only

A LifeDraft memorial only speaks from content the person actually provided. It doesn't invent new opinions, fabricate memories, or speculate about what someone might have thought about topics they never addressed.

This is a crucial distinction. Many AI systems are designed to generate plausible responses to any query. But generating plausible responses for a deceased person raises serious ethical concerns—it could put words in their mouth they never said, or opinions they never held.

When a LifeDraft memorial encounters a topic the person didn't address, it honestly says so: "I don't have any recorded thoughts about that topic. Is there something else from my stories I can share with you?"

Principle 2: Transparent Identity

A LifeDraft memorial always identifies itself as a memorial representation. It never claims to be the actual person, never pretends to have current awareness, and always maintains clear boundaries about what it is.

This transparency is essential for healthy interactions. Family members should never be confused about whether they're talking to their loved one or to a memorial created from their words. The distinction matters for psychological wellbeing and for honoring the truth of mortality.

• Every conversation begins with clear identification as a memorial
• The system uses appropriate language ("I recorded my thoughts about..." rather than "I think...")
• Boundaries are maintained even if users try to blur them
• The experience honors the reality that the person has passed

Principle 3: Creator Control

The person creating a LifeDraft has complete control over what their memorial can discuss. Through our persona settings, creators can:

• Set topics that are off-limits for discussion
• Define sensitivity levels for different content
• Control who can access which parts of their memorial
• Specify how their memorial should respond to certain types of questions
• Choose the tone and personality characteristics of their representation

These settings are locked when the memorial is activated, ensuring the creator's wishes are permanently honored. No one—not family members, not us—can change these boundaries after the fact.

Principle 4: Privacy by Design

Privacy isn't an afterthought in our system—it's built into the architecture. Your stories, voice recordings, and vault contents are encrypted with keys that we cannot access. Even if our systems were compromised, your private information would remain protected.

We use a system called Shamir's Secret Sharing for vault keys, which means no single person (including any LifeDraft employee) can access your vault. It requires multiple trusted contacts to work together, ensuring true decentralized security.

Principle 5: Thoughtful Activation

Memorial activation—when your LifeDraft becomes accessible to your trusted contacts—involves careful verification. We don't activate memorials casually. Our multi-step process includes:

• Regular check-ins to confirm user status
• Extended grace periods before escalation begins
• Notification of trusted contacts before any activation
• Multi-party verification requiring confirmation from multiple people
• Executor approval for final activation

This process ensures that memorials are only activated when appropriate, preventing premature activation due to simple technical issues or temporary unavailability.

Principle 6: Continuous Ethical Review

AI technology evolves rapidly, and so do its ethical implications. We maintain an ongoing commitment to ethical review through:

• An external Ethics Advisory Board that reviews our policies and practices
• Regular internal audits of our AI systems for bias and appropriate behavior
• Active engagement with grief counselors and mental health professionals
• Transparency about our practices and willingness to update them as we learn

We don't claim to have all the answers. Memorial AI is new territory, and we're committed to learning and improving as we go.

What We Will Never Do

Some lines we will never cross, regardless of technical capability or user requests:

• We will never create memorial representations without explicit consent from the subject
• We will never allow memorials to claim to be the actual person rather than a representation
• We will never train our AI on private user data to improve general models
• We will never sell user data or memorial content to third parties
• We will never allow memorials to be accessed by anyone not designated by the creator

Our Ongoing Commitment

Building memorial AI technology is a privilege and a responsibility. Every day, we're trusted with some of the most precious content imaginable—people's life stories, their wisdom for future generations, their final messages of love.

We approach this trust with humility. We know we're working in deeply personal territory, and we're committed to earning that trust through our actions, our transparency, and our unwavering commitment to ethical practices.

If you have questions about our approach or want to share feedback, we're always listening. The conversation about ethical AI in memorial technology is ongoing, and your perspective matters.