Anthropic’s consultation with Christian leaders to shape the ethical framework of its Claude chatbot has raised concerns about religious bias in artificial intelligence increasingly used in schools.
Anthropic hosted Christian leaders in March to discuss Claude’s moral formation, according to Brian Patrick Green, a director of technology ethics at Santa Clara University, who attended the meeting and described the discussions to The Washington Post. Claude powers educational tools such as MagicSchool, which the Sequoia Union High School District (SUHSD) adopted at its March 4 board meeting for educators and students. While the current platform remains unaffected, Green’s account of Anthropic’s consultation has led educators to reexamine the principles guiding the software students use daily.
“We all have values. We all have upbringings. We all have community influences,” said Jadie Sun, a computer science teacher at Carlmont. “If they affect chatbots, and that’s used to educate students, there can be bias, and there can be a narrowing of people’s points of view and freedom of thought.”
As AI tools become more widely used in classrooms, the values embedded in those systems are under scrutiny, raising questions about neutrality in public education.
Anthropic has positioned itself as an AI ethics leader through its Constitutional AI approach, which trains Claude on guiding principles to make it safe and helpful. The company further reinforces this stance by refusing to loosen safeguards, even for national defense projects.
However, critics argue that Anthropic’s latest consultation with Christian leaders risks introducing bias by excluding other religious perspectives.
“You should get as many views as possible because AI should be a think tank of all different views, and the more diverse, the more we know, the more informed everybody is,” Sun said.
As AI firms prioritize ethics, classrooms and organizations are debating what these principles should mean for students.
The College Board is a prime example of incorporating ethics-related learning goals into classes such as Advanced Placement (AP) computer science courses. AP Computer Science A (AP CSA) focuses primarily on Java programming and introduces the ethical and social implications of computing, while AP Computer Science Principles (AP CSP), which Carlmont does not offer, emphasizes responsible computing.
However, some AP CSA students say there is limited instruction on ethics, particularly as AI becomes more central to education and future careers.
“I think the AP CSA class focuses a lot more on technical skills. We learn a lot more about logic and how to think about problem solving,” said Carlmont sophomore Hila Dremer.
Sun echoed those sentiments, noting that ethics receives limited emphasis in the course.
“It is part of the AP CSA curriculum, but it is not tested, and when things aren’t tested, students tend to give it less priority,” Sun said.
Despite the curriculum gaps, students are proactively exploring their interests in ethics, according to Sun, who advises Carlmont’s student-led AI and Machine Learning Club.
“In that club, they look into machine learning and how to improve how AI systems learn,” Sun said.
Broader efforts to incorporate ethics-based approaches into AI education are emerging. In April, the University of Notre Dame’s Institute for Ethics and the Common Good (ECG) awarded up to $1 million in grants to develop a K-12 curriculum grounded in the faith-based DELTA framework. This initiative raises questions about theology-influenced moral frameworks in secular public schools.
“We have actually not encountered any strong pushback to the faith angle that DELTA takes. The core values of the framework are widely agreed upon by everyone we’ve shared it with,” said ECG Program Manager Julia French.
French said feedback has focused instead on implementation.
“The bigger pushback we’ve received is that the framework doesn’t explicitly propose any concrete solutions,” French said. “It instead provides a lens through which you can wrestle with AI problems and that key flexibility that’s needed for both a school in Northern California to adopt it and a rural school in Louisiana to pick it up.”
Whether faith-influenced frameworks might fit within secular public schools remains unclear.
“As we would with any resource or curriculum, the district would approach implementation of AI platforms by emphasizing transparency, ongoing evaluation, and alignment with educational goals,” said SUHSD Director of Instructional Technology Barbara Reklis.
The convergence of faith and AI is not new. It began in February 2020 with the Vatican-led Rome Call for AI Ethics, which outlined a human-centered approach to AI. The initiative has since garnered more than 80 signatories from religious leaders, technology companies, and major corporations.
Following the partnership with MagicSchool, SUHSD plans to establish guardrails to encourage safe and responsible AI use, in accordance with California Department of Education guidelines. The district also hosted an AI listening session on April 28 to gather input from students, teachers, staff, parents, and community members on priorities and concerns about AI integration in education.
“AI tools play a role in our education, and it’s important that they come from companies we can trust,” Dremer said.
As AI integrates further into daily life, the debate over who gets to define its moral compass and whether those definitions belong in a classroom is just beginning.
“Our top priority is to ensure our students have what they need to achieve academically while maintaining safety,” Reklis said.
