What happens when AI starts targeting Christians?

By Andrea Picciotti-Bayer, Voices Contributor Monday, June 22, 2026

This month, five FBI employees were fired for their roles in producing a 2023 internal memo suggesting that Catholic congregations could serve as venues for intelligence gathering on extremist groups.

An internal review found that those involved had failed to follow analytical standards and improperly connected religious beliefs to violent extremist ideology without sufficient evidence. The memo stemmed in part from reliance on the discredited Southern Poverty Law Center and the religious affiliation of a single law enforcement target who had identified himself as a traditionalist Catholic. 

The firings close one chapter. They should open another. 

The Richmond memo was produced the old-fashioned way: human analysts, working from ideologically skewed sources, drew conclusions about a religious community that the evidence did not support. The damage was real but limited in scope. Now consider the same dynamic accelerated and automated by AI-driven programs that can cull social media, map worship patterns, correlate data points across platforms, and flag behavioral “anomalies” at scale — continuously, without a supervisor, a congressional hearing, or an inspector general review. What took a team of analysts months to produce badly could be replicated instantly. The biases embedded in training data do not disappear; they scale. 

To see where that logic leads when left unchecked, look to China. Authorities have installed cameras at the entrances of churches and in some cases inside sanctuaries, tied to facial recognition systems that identify worshippers and feed information directly into state databases. Churches that refuse installation have been fined, raided, and shuttered. The Zion Church in Beijing — one of the capital’s largest congregations — was closed after declining government orders to surveil its own sanctuary. In 2025, authorities raided Zion again and arrested its pastor on charges of illegally disseminating religious information over the internet.  

China's facial recognition network is being integrated with its Social Credit System, which grades the ideological loyalty of citizens, giving the state both the infrastructure and legal framework to pressure religious minorities in ways that have no historical precedent. The U.S. Commission on International Religious Freedom’s 2025 annual report documents how China deploys AI, facial recognition, voice recognition, and big data analytics to monitor targeted religious groups, treating religious belief itself as a threat to state control. International religious freedom expert Gia Chacon has a chilling observation: “First deployed against Uyghur Muslims, the same system has been used to monitor Christian communities.”
Experts also warn that Beijing’s model is not staying within China’s borders but instead is a testing ground for AI-enhanced social control being exported to other authoritarian regimes. What begins as a domestic instrument of repression becomes a global template. 

Here is where an unexpected voice enters the conversation. When Pope Leo XIV published Magnifica Humanitas last month, most commentary focused on its sweeping call to regulate artificial intelligence and its warnings about algorithmic power concentration. Less noticed was its direct bearing on religious freedom and on the specific policy failures the Richmond memo and the Zion Church case expose. 

Pope Leo explicitly affirmed that religious freedom is a fundamental right grounded in human dignity that must be guaranteed by law. He argued that automated decisions must be explainable and open to challenge and warned against social control made possible by the massive collection of data and algorithmic systems. For a secular policy audience, the significance of the encyclical shouldn’t be ignored. The head of the world’s largest religious institution has identified AI-enabled religious surveillance as a live threat and called for legal structures to constrain it. That is not a position without policy weight. 

AI systems already govern hiring, lending, healthcare triage, educational admissions, and access to government services. When those systems are trained on biased data or flag religious practice as “anomalous,” discrimination becomes invisible by design. There is no human decision to appeal, no supervisor to override the outcome. This is why the demand for transparency and human oversight of automated decisions, including the right to explanation and challenge, is, at its core, a religious freedom argument as much as a technological one. Religious accommodation claims involving AI decision-making in employment, healthcare, and education are inevitable.  

Religious surveillance is a civil liberties problem. It is also a genuinely new one that is novel enough that lawmakers and courts are still developing the vocabulary to describe it, let alone the tools to address it. As legislatures at every level scramble to regulate AI, the religious liberty dimension of that project remains largely unnamed. Naming it clearly and building it into the frameworks now taking shape is an opportunity that will not stay open indefinitely.