Three months after Pope Leo XIV’s encyclical Magnifica humanitas called for vigilance and transparency in governing the use of artificial intelligence, individual dioceses in the United States take varied approaches to the use of AI by employees.
Few dioceses have publicly available AI policies published online, while some say they are currently working on creating policies.
Those that do have published AI policies often include statements about human dignity and transparency. But what is actually permitted as far as the use of AI by diocesan employees varies from one diocese to the next.
Some dioceses allow the use of generative AI to draft emails, outlines and school materials, while others limit it to more narrow roles, including summarizing, editing and proofreading. In general, dioceses require human oversight and review of AI-generated content. They also prohibit entering confidential information into AI systems. However, few diocesan policies reviewed by The Pillar require a label indicating that AI was used to produce content.
The Vatican City State introduced its own AI guidelines, which took effect at the start of 2025, a year-and-a-half before Pope Leo’s encyclical on the subject, and were meant to set the framework for eventual laws on the subject.
The Vatican City State guidelines say that AI must be used in ways that respect “the security of the Vatican City State, the protection and confidentiality of personal data, non-discrimination of human beings, economic sustainability and care of Creation.”
They repeatedly emphasize the need for “preserving respect for human autonomy and decision-making power” and ensuring that AI systems are supporting humans rather than replacing their judgment.
The Vatican guidelines call for transparency in the use of AI, and say results must be monitored to ensure correct and reliable results.
Data processing and presenting information must be done in a way that respects both truthfulness or freedom of expression, the Vatican guidelines say. Copyright must always be respected.
In judicial decisions, AI may only be used “for the organization and simplification of judicial work, as well as for research,” but not for the interpretation of the law, which may only be done by a magistrate.
The guidelines prohibit the use of artificial intelligence in ways that could cause physical or psychological harm through subliminal manipulation, or in ways that exclude people from disabilities with accessing the features of artificial intelligence, as well as the use of data processing in ways that violate human dignity and create social inequalities.
The Vatican also created a Commission on Artificial Intelligence to monitor the evolving use of artificial intelligence and eventually propose laws for the city state.
The Diocese of Biloxi enacted a policy on the use of AI at the start of 2026.
The policy permits diocesan employees to use AI only with express authorization from their supervisor.
“AI may, in certain limited circumstances, be utilized in an authorized manner as a research tool, but not a replacement for human discretion and intellect,” the policy says.
Generative AI may be used “to enhance efficiency in proofreading, summarization, and conducting preliminary research,” as well as in preparing teaching instruction, calculations, and data analysis, the policy states. All AI-generated content must be reviewed and verified by a person.
AI may not be used to generate the first draft of a text, the Biloxi policy says. And work created with the help of AI must be clearly labeled to show how AI was used.
AI systems should never be given access to confidential information, including financial data, donor information, content from confidential meetings, access to emails or calendars, and pictures or information about minors, the policy says.
In addition, students may not use AI to answer questions without permission, and teachers may not use AI to determine if a work submitted by a student was AI-generated.
The Diocese of Paterson has also released a policy on AI use. It states that individuals must review AI-generated content for accuracy, and AI should not be permitted to make “final decisions without human oversight.”
The policy prohibits entering confidential meeting information into AI without clearance. It also prohibits submitting information into AI systems in ways that could violate clergy confidentiality.
AI may not be used to impersonate people, edit pictures of minors, or analyze biometric data without consent, according to the diocesan policy. AI systems may not be permitted to post on diocesan social media accounts without human oversight.
AI should not be for legal, medical, or counseling advice without “proper professional involvement,” the policy says.
“All AI outputs should be stored and archived according to diocesan record retention policies,” it adds.
The Archdiocese of Los Angeles says in its brief policy that generative AI “may be used to assist in work and ministry but must always be evaluated by the user for accuracy and shall not be a replacement for human discernment.”
Employees must use an archdiocesan email to log into their generative AI tool. They must avoid entering confidential information, impersonating other people, engaging in illegal activities, violating privacy and copyright laws, interfering with their work or other employees’ work, and breaking other archdiocesan rules.
Students in archdiocesan schools may generally use AI, with teacher permission and guidance, “for homework assistance, research, or writing support as a supplement to active engagement.” Work must be properly cited as using AI.
Teachers may use AI for lesson planning and grading, as long as student identifiers are not used. They may also use AI faith resources, such as Magisterium AI.
The Diocese of Orange is governed by what it calls a “living document” on AI, overseen by the Diocesan AI Council.
The diocesan policy allows AI use for a wide variety of tasks, including streamlining administrative work and communication, creating catechesis materials, and liturgical planning.
AI may be used in the diocese to create bulletins, podcasts, chatbots, and children’s videos, as well as to create digital books, educational games, and tests in Catholic schools.
The diocesan policy prohibits using AI in ways that violate privacy and personal data, or that reinforce social, racial, or economic inequality.
“AI must never replace human interaction in areas requiring empathy, spiritual guidance or moral discernment, such as pastoral care, spiritual direction or sacramental preparation,” the policy says.
AI systems may not be used to “make moral, ethical, or theological decisions,” it adds.
The Archdiocese of Denver prohibits AI tools that require the use of archdiocesan data for training. It specifically permits the use of ChatGPT and Microsoft Co-Pilot Transcription.
The archdiocese policy explicitly permits the use of AI for brainstorming, summarizing research, drafting emails and lesson plans, designing presentations, data analysis, and creating educational materials.
It prohibits using AI without human oversight and entering confidential information into AI systems.
It also prohibits “relying on AI for moral, pastoral, or theological decisions” and “using AI to replace personal interaction in ministry, teaching, or pastoral care.”