Verghese V Joseph –
Synderesis, an independent Polish AI platform, has built a model trained to reason in line with Catholic values—human dignity, Church teaching, privacy, and human responsibility. For Indian Catholics, it raises a timely question: can AI support parishes, schools, charities, and everyday work while preserving conscience, pastoral judgment, and the primacy of the human person?
Co-founders Oscar Andersson and Dr. Hasse Hällström discuss how a “Catholic by design” AI could help with writing, research, administration, and formation—without becoming a substitute for priests, educators, community discernment, or sacramental life.
They also address affordability, multilingual needs, data privacy, India’s diverse Catholic traditions, and why AI must remain a tool in service of the common good. Synderesis says its model embeds Catholic values during training rather than merely adding religious instructions to a general-purpose system; it also stresses that it has no official Church endorsement and that users remain responsible for decisions. Excerpts:
What makes Synderesis “Catholic by design”?
We founded Synderesis because we wanted Catholics to have a real choice in AI, including a choice about the values built into the underlying model.
We have developed what we believe is the first publicly available AI model specifically retrained around Catholic values. We start with open-weight models and retrain them using examples informed by Catholic teaching. That is different from taking a general-purpose model and adding Church documents or religious instructions without changing the model itself.
We are proud of that technical achievement, but its purpose is practical. We want Catholics, schools, charities and businesses to use AI for writing, research and everyday work without setting aside their convictions. The model can assist; people still review its answers and make the decisions.
What Catholic principles should guide AI adoption in India?
First, access. AI should be useful to a small Catholic school or charity, not only to a wealthy business. Affordability matters, especially when a tool could save staff time for the people they serve.
Working with our own retrained models rather than simply putting a Catholic interface over someone else’s AI also gives us more control over costs and pricing. It creates room to offer competitive API prices, which matters for developers, schools and charities in India that want to reach more people on limited budgets.
Second, ask what moral assumptions the developers built into the system. The AllFaith Benchmark, developed by researchers from universities including Notre Dame and Baylor, found that mainstream models often omitted religious perspectives from everyday ethical questions and showed measurable differences in advice about religious conversion. That shows why Catholics should not simply rely on foundation models trained around questionable assumptions, and why having a model grounded in Catholic teaching matters. For us, the essential test is whether AI helps people do good work while leaving responsibility with them.
Could Synderesis encourage interreligious dialogue?
We hope so. A clear Catholic identity should not prevent respect for other beliefs; it should make honest dialogue possible.
Oscar is Swedish and I (Hasse) is Finnish, and we have both lived in Poland. We know what it is like to move between largely secular Nordic societies and a country where Catholic life is much more visible. That experience matters when building technology for people with different beliefs.
Nostra Aetate is an important guide for us. In India, where Latin, Eastern Catholic and many other religious communities live alongside one another, we would want Synderesis to encourage understanding rather than flatten those differences.
What can India learn from Europe’s AI approach?
Europe has taken privacy and individual rights seriously through the GDPR and the EU AI Act. But regulation alone does not solve a second problem: dependence on a handful of technology providers.
If schools, charities and businesses rely on models controlled elsewhere, changes to prices, access or policies can affect their work overnight. That is why we care about open-weight models and more independent deployment options. Europe needs them, but so does India.
For organisations handling sensitive information, we would like to make private-cloud and locally hosted approaches more accessible.
How can Synderesis serve India’s diverse Catholic traditions?
We should start by listening. India’s Latin, Syro-Malabar and Syro-Malankara communities have rich and distinct traditions. We would not want to treat one European experience of Catholic life as universal.
We would welcome conversations with Indian dioceses, universities, schools and businesses about where AI could genuinely help, and what would need to change for regional languages and local needs. We would rather build with Indian Catholics than simply market a finished product to them.
How do you interpret Pope Leo XIV’s statement that “Technology is never neutral”?
For us, it means looking beyond what an AI tool can do and asking what human choices shaped it. As Magnifica Humanitas (paragraph 99) states, AI systems do not undergo experiences, lack a moral conscience and cannot bear responsibility for consequences. Data, training choices and rules set by developers determine how a model responds.
One revealing example is Anthropic’s published constitution for Claude, which treats whether the model is a moral patient as an open question and describes it as a virtuous agent. From a Catholic perspective, attributing moral status or agency to machines blurs the unique dignity of the human person.
In paragraph 107, Pope Leo XIV also warns that a more moral AI is not enough if that morality is determined by a few. Catholics should have a foundation model grounded in their faith rather than questionable moral premises, which is why we trained Synderesis around Catholic teaching.
Which Catholic principles shape Synderesis most directly?
Human dignity comes first, together with the common good, solidarity, subsidiarity and personal responsibility.
I think the dignity of work makes this concrete. A teacher should be able to spend more time with students, not less. A charity should have better tools to serve people, without handing over its judgment to software. A small business owner should be able to work more effectively without becoming dependent on a system they cannot question. While some major AI charters define success as outperforming humans at most economically valuable work, Magnifica Humanitas (paragraph 152) insists on protecting human work. We aim for useful assistance that strengthens human capabilities rather than treating people as replaceable.
How do you respond to concerns about lacking Church endorsement?
We do not have official Church endorsement, and we have never claimed to. We started Synderesis in August and have only recently begun to introduce it publicly. We would welcome dialogue with Church institutions and, in time, their endorsement.
Our goal is not to act as a doctrinal authority or build another tool focused mainly on looking up Church teaching. We want Synderesis to be the AI assistant Catholics turn to in everyday life and work, whether they are writing, researching, teaching or running an organisation. Catholic teaching guides our approach, but we do not speak for the Church, and our model can still make mistakes.
How does Synderesis approach sensitive pastoral questions?
Synderesis can help people find and understand Catholic teaching, but pastoral questions often require knowledge of a particular person and circumstances.
Where that is the case, the right answer may be to consult a priest or another appropriate person. Our main purpose is different: helping Catholics with writing, research and everyday work. We want to support people’s judgment, not take it over.
What data does Synderesis collect, and what control do users have?
Privacy, security and regulatory compliance were among the reasons we founded Synderesis. We believe Catholic institutions should be able to use powerful AI without giving up control of sensitive information. That means collecting and retaining as little data as we genuinely need, protecting what we do handle, and giving users meaningful control over their information.
We collect the account and service information necessary to operate Synderesis, and we do not sell personal data. In our browser application, recent chat history is stored locally on the user’s device. Users can delete chats, use Ghost mode so a conversation is not saved in browser history, or delete their accounts.
We are building Synderesis around strong privacy, safety and data-protection standards, including the GDPR. We use safeguards such as encrypted connections and protected account credentials. We want privacy and compliance to be genuine strengths of Synderesis, not just promises in marketing.
For Catholic schools, charities and other organisations handling sensitive information, we are working towards private-cloud and on-premise deployments. Our aim is to give institutions greater control over their AI infrastructure and the information entrusted to them.