AI and religion. When synthesized voices reach the pulpit.

A field note on a use we did not predict.

Context

When we launched, we expected the detector to be used mostly for fraud and journalism. We did not expect requests from religious organizations. We started receiving them in mid-2024.

What we have seen

Three patterns. First, synthesized voices of deceased religious leaders used in sermons or memorial events, sometimes attributed clearly and sometimes not. Second, AI-generated voices delivering recorded prayer or scripture in pre-recorded services. Third, allegedly miraculous recordings (voices of the departed, voices in the wind, voices in dreams) submitted to the detector by skeptics and believers alike.

We treat all three the same way. We return a probability. We do not opine on the underlying question.

The ethical questions

Three distinct ones. The first is consent: did the person whose voice was cloned consent? For deceased individuals, who can consent on their behalf? The second is disclosure: is the congregation told the voice is synthesized? The third is fitness for purpose: are there contexts where synthesized voice is inappropriate even with consent?

These are not our questions to answer. They belong to the communities involved.

Our position

We provide a verdict on the audio. We do not provide a verdict on the practice. If a community wants to use synthesized voice with full disclosure, that is not our business. If a community wants to verify a recording attributed to a specific person, that is exactly our business.

The detector measures audio. It does not adjudicate practice. We try to keep that line clear.