Use of generative artificial intelligence (AI) in writing sermons has sparked a wide range of responses among Anglican clergy, from enthusiastic embrace of the technology to pledges totally opposing its use.
More than 150 preachers and church leaders, including about a dozen Canadian Anglicans, have signed the Saint Dunstan Pledge for Preachers, committing not to use generative AI to write sermons, homilies or reflections. The Rev. Emily Garcia, an Episcopal priest in the diocese of Massachusetts, created the online pledge on May 19—the feast of St. Dunstan, patron saint of metal workers and other craftspeople.
The Saint Dunstan Pledge declares that signatories will not use generative AI tools such as ChatGPT, Claude, Copilot or Gemini at any point in the brainstorming, planning, drafting or writing of their sermons. It clarifies that this pledge does not prevent them from using non-generative AI for tasks such as drafting voice memos, converting text to spoken audio, correcting spelling and grammar or doing basic Google searches without AI.
Canon Maggie Helwig, priest in charge at the Church of St. Stephen-in-the-Fields in Toronto and a published writer of poetry, fiction and non-fiction, says she signed the Saint Dunstan Pledge due to a broad range of concerns. Helwig points to the ecological impact of AI data centres, such as their vast energy and water consumption. She also worries about intellectual property theft in the development of large language models that produce output based on the work of artists and writers who receive no compensation.
Helwig describes generative AI as “burning down human civilization to enable mass consumption of predictive text from the plagiarism machine.” The Rev. Helen Dunn, rector of St. Clement’s Anglican Church in Vancouver, says she signed the Saint Dunstan Pledge as an act of “solidarity with artists and creatives who are slowly losing their craft to AI.”
Preaching, Helwig says, is an “incarnational act” in which humans interpret ancient texts to their communities, based on their own struggle and relationship to the texts. She says teaching and celebrating the sacraments was the first responsibility placed upon her at her ordination. “Why would I want a jumped-up predictive text model interfering with that sacred process?”
The Rev. Grace Burson, incumbent at the Church of All Saints By the Lake in Dorval, Que. and a writer who also signed the Saint Dunstan Pledge, calls use of generative AI for sermons “an abdication of our vocation as preachers and writers and thinkers.”
“We made promises at our ordination to use our gifts, our brains, our hearts to serve our people,” Burson says. “The sermon is the primary place where I encounter my congregation theologically, where we have that conversation about scripture and about how God is moving in our lives. The idea of just typing a prompt into a search bar and having it spit something out for me … my skin crawls when I think about doing that.”
The Rev. Tay Moss—honorary associate at St. John the Baptist Norway Anglican Church in Toronto, director of the online church learning platform CHURCHx and CEO of consulting service Leporia Software, which describes itself as providing “digital tools for faith communities,” including AI—offers a different perspective.
“AI needs to be understood as another tool in the toolkit that pastors and priests use for sermon writing,” Moss says. He compares generative AI to other online resources such as commentaries, podcasts or archived sermons that a preacher might consult in writing a reflection. AI can offer suggestions to help a preacher improve the flow of their argument, Moss says, or tailor their sermon to a specific audience.
“There’s a spectrum here … That idea of people sitting there and just lazily writing a prompt and having it generate the whole sermon—yeah, that’s not a good idea, but I think that’s evident to everybody,” Moss says. He describes the Saint Dunstan Pledge as an “all-or-nothing, reactionary approach to something that requires a lot more theological nuance and discussion.”
Opposition to generative AI, Moss says, often comes from fear or anxiety about larger societal issues such as climate change and mass surveillance. Comparing AI to previous new technologies such as cars that were eventually regulated, Moss says AI and data centre construction should likewise be regulated.
In June, Council of General Synod (CoGS) passed a resolution directing the offices of the primate and general secretary to establish a working group to draft terms of reference for an AI ethics task force, with a mandate “to establish guiding virtues” to help the Anglican Church of Canada “negotiate faithful engagement” with generative AI technologies. Moss says he hopes the AI ethics task force will establish “guardrails and principles” rather than strict rules.
Bishop of Central Newfoundland John Watton, a member of CoGS, says people should treat AI as a research assistant. Clergy should give credit if they choose to use AI in writing sermons, Watton says, as they would any other source—and must be aware that AI can generate information that is false, biased or lacking context.
“What we say in our diocese is that we hold each other lovingly accountable … We’re saying to our clergy, if you use [AI], learn how to use it properly,” Watton says. “If you quote AI, then you give it credit and you let people know that this is where you’re working from. I think that’s fair because you actually are using someone else’s idea and flow of thought.”
The Rev. Greg Little, associate pastor at St. John the Evangelist Anglican Church in Strathroy, Ont., says AI has developed rapidly since he wrote a blog post in January 2024 reflecting on the technology’s use in sermons. He describes preaching as “a special calling” that “really has to be authentic and it has to come from you.”
Little says he is not opposed to preachers using AI to write sermons if they make clear that they have used AI, even as a source of information.
“There’s got to be a clear distinction between something that AI produces and something that a human produces,” Little says—something he acknowledges could become “harder and harder to distinguish” in the future as AI technologies continue to advance.
