AI is here to stay, and the imperative is to engage with it, and engage responsibly, participants concluded at the online discussion “Responsible AI for Faith-Based Organizations” hosted by WACC’s European regional association (WACC Europe) on December 17.

The event featured presentations from Diakonie Deutschland and WACC member the Church of Sweden, both of which have taken concrete steps in developing and implementing guidelines for using artificial intelligence within their respective organizations.

The two faith-based organizations highlighted the opportunities AI presents to assist with particular tasks but also underscored the critical need to address the associated risks and ethical concerns.

Guidelines to avert “shadow AI”

Arni Svanur Danielsson, head of communications at WACC member The Lutheran World Federation, opened the discussion by pointing to recent studies showing that, within companies that banned the use of AI or did not have guidelines, 40–60% of their employees were still using AI in their work.

He noted that this “shadow AI” carries the highest risk because the employees would most likely be using free versions which do not protect data, and staff of organizations without AI policies will often have the least amount of training in how to use such technology most usefully and ethically.

“Guidelines are a first step because they provide a framework,” Danielsson said, but added that there needs to be education and ongoing discussion in organizations to foster a culture where AI is used knowingly, responsibly, and transparently.

Overall, “AI is like an intern or an assistant – they need guidance and oversight to be useful,” he said.

Policies that maintain data privacy

Evelyn Moeck shared the experiences of Diakonie Deutschland, which has about 850 employees. Soon after the launch of ChatGPT two years ago, the organization decided that there needed to be organizational responsibility to ensure ethical, safe, and legal use of AI, as well as to secure potential gains in productivity.

The process involved a focal group with broad participation from all departments which met weekly for seven weeks to develop the guidelines, along with a broader discussion to allow all employees to participate. The organization conducted tests in different departments as well as involved contract partners and alliances in the process.

Moeck outlined key takeaways from Diakonie Deutschland’s experience:

• Developing guidelines takes time and requires consideration of all stakeholder groups.
• Acceptance and accountability are key, with a recognition that everyone is in a learning journey.
• Testing is crucial.
• Professional help is needed to understand the technology.

A core guiding principle in Diakonie Deutschland’s guidelines is “Don’t give any data into an AI system which you wouldn’t share with anyone out on the street”. Moeck said this helps to maintain high standards for data privacy.

Investing in development of AI policies

Ulrika Ekberg, from the Church of Sweden, shared her organization’s experience in developing an AI policy which was adopted in June 2024. The policy was a response to a clear need for guidance on the responsible use of AI within Church of Sweden, she said, noting that employees were already exploring how AI could assist with their tasks as an “extra colleague”, leading to AI discussions in various situations.

The policy is relatively permissive, with a positive view of AI and emphasizing responsibility and transparency. It also aligns with EU guidelines promoting the responsible development of AI and prioritizes efficiency while following laws and regulations.

Ekberg noted some key elements of Church of Sweden’s policy:

• Restricting the use of free applications
• Requiring the approval of data owners
• Assessing risks from operational, financial, strategic, ethical, adherence, and integrity perspectives
• Revising the guidelines every six months to keep pace with the rapid development of AI

Currently, Church of Sweden is conducting a pilot study of Microsoft 365 Copilot with 50 participants. The aim is to explore how the tool can benefit daily work, while mitigating the risks and challenges related to data security, over-dependency, and the potential for biases.

Among the risks, Ekberg noted that, contrary to many people believing AI would save money, the technology is actually quite costly to initiate and maintain although with appropriate education and application it can streamline routine tasks and allow organizations and businesses to concentrate more on their core mission.

Both presentations highlighted the need to give time to the process of developing, testing, implementing, and updating guidelines.

Non-engagement with AI is not an option

“How AI sees us” was raised in the discussion, as one participant noted that when they did trial queries related to churches, they found they could just be associated with “old buildings”. Others experienced realistic-looking references that turned out to be fake.

“AI gives the most plausible answer, not necessarily the correct answer,” Moeck noted. She said that education about what AI can and cannot do is important, as it helps everyone be more realistic.

All participants also acknowledged there must always be a human in the loop. As one participant said, “What we produce is our own responsibility, not just a copy-and-paste from an AI chat.”

What was clear, was that not getting involved is not an option. Moeck observed how fast AI is developing, and how scary some of the developments and uses are. She expressed concern that faith-based organizations are moving too slowly to address it in a good way. The bottom line: “AI is here, and it will not go away.”

“Responsible AI for Faith-Based Organizations” was the first in a series of monthly discussion that WACC Europe has launched to explore key topics facing communicators. The online events take place every third Tuesday from 10-11am Central European Time. The next topic will be “Fragmentation and Polarization: The Media Space”, on January 21.