66 Indigenous Family Service Centers and 222 Social Workers Enter Disaster-Response Training: Five Cards for Culturally Responsive Services

From October 12 to 14, 2026, Taiwan's Council of Indigenous Peoples, Taiwan is holding the second in-service training session for social workers from Indigenous Family Service Centers. The session brings together 222 frontline social workers from 66 centers nationwide. Its curriculum links case discussion, Indigenous social-work theory, welfare policy, culturally responsive practice, work with vulnerable families, community resource networks, and service delivery during disasters.

That combination matters. The hard part of public service in Indigenous regions is rarely whether one more data field exists. It is whether information remains connected to family relationships, cultural context, geographic constraints, service responsibility, and the person's own choices. One household may face income, caregiving, medical transport, schooling, housing, interpretation, or disaster-recovery needs at the same time. A system that reduces those circumstances to one risk score can miss who has authority, which local relationship is trusted, and what requires immediate human action.

The 2027 implementation plan provides the institutional picture. As of June 2026, 13 local governments operated 66 centers with 222 Indigenous social workers. Taiwan's 55 officially designated Indigenous townships and districts are prioritized for center coverage. Services include economic and daily-life support, family relationships, child and adult care, personal adjustment, community work, welfare coordination, population-needs assessment, and rapid care and recovery support after major incidents and disasters.

The economic significance is practical. Earlier identification, coordinated transport and care, and closed-loop referrals can reduce the time, travel, and information costs carried by families. During disasters, a center may become the node that connects formal warnings, household needs, local networks, and long-term recovery.

Five culturally responsive service cards can make that node more reliable. The cards do not turn professional judgment into a checklist, and they do not authorize AI to decide. They keep necessary facts, responsibility, consent, and next actions from disappearing between organizations.

The first is a contact and needs card. It records how the person wishes to be addressed, preferred language, safe contact time and channel, the most urgent need, any immediate safety concern, and a promised response time. When information comes from a relative, a community partner, or another agency, the card identifies that source instead of presenting a report as confirmed by the person.

Only necessary data should be collected. Identity, health, family, financial, and contact information may all be sensitive. Each field needs a defined purpose, access role, retention period, correction path, and deletion procedure. An emergency contact list must not quietly become a general outreach list.

The second is a cultural context and consent card. It may record language and interpretation needs, people the family wants involved, times or settings to avoid, the specific information that may be shared, and the form of support the person prefers. These fields come from dialogue; they must not be inferred from a name, address, or model prediction.

Cultural responsiveness is not stereotyping. Families from the same people or community may differ in kin participation, ceremonies, gender roles, care arrangements, and trust in outside institutions. The card must preserve states such as unknown, needs confirmation, and not consented. Consent should be granular, time-limited, and revocable.

The third is a coordination and referral card. Every need is connected to an accountable lead, supporting parties, a deadline, and a feedback method. Emergency support, medical transport, childcare, legal advice, employment, and long-term care may have different providers, but the family still needs one person who can explain the whole journey. A referral is complete only when the receiving service confirms intake, the family understands the next step, language and transport barriers are addressed, and an alternative exists if the handoff fails.

The fourth is a disaster and service-continuity card. Before an emergency it identifies mobility, medication, child or elder care, communications gaps, transport alternatives, livestock or pets, and trusted local support. During an event it adds the official alert version, successful contact, current location, requested service, and next check-in time. The card supports care and coordination; it does not issue evacuation, return, or safety decisions.

Taiwan's Disaster Prevention and Protection Act assigns governments duties for warning delivery, evacuation, alert zones, temporary shelter, social assistance, and special protection for vulnerable groups. Family service centers can translate those duties into a household journey: who received the message, who needs Indigenous-language or in-person explanation, who cannot act without transport or replacement care, and what remains unresolved after an alert is lifted.

The fifth is a data and decision audit card. Any AI-generated summary, referral suggestion, work queue, or multilingual draft should retain the input version, output time, human reviewer, edits, final decision, and notification result. Missing data must remain visible. Conflicting records must be presented side by side instead of being concealed by a fluent sentence.

AI can organize meeting actions, flag overdue replies, detect duplicate service activity, draft plain-language or multilingual text for review, and identify cases with no owner or next step. It must not infer cultural needs from identity, rank families through opaque risk predictions, expand data sharing, deny services, or issue disaster-safety conclusions.

Taiwan's Social Safety Net 2.0 is centered on families and communities and includes cross-disciplinary training and responsible technology use. The Personal Data Protection Act requires collection and use to remain within a necessary, specific purpose. Together, those standards mean success should not be measured by faster form completion alone. Families should travel less, repeat painful stories less often, reach the accountable worker sooner, and retain practical routes to access, correction, restriction, and withdrawal.

A one-month pilot in an Indigenous township can test the five cards with one routine service and one disaster exercise. Families, social workers, community partners, and local government can check closed-loop referrals, Indigenous-language and offline access, data minimization, and complete human review of AI drafts. Useful indicators include first-response time, successful handoffs, repeated storytelling, overdue actions, consent-withdrawal time, and correction time.

The October 12 training places cultural response, vulnerable families, community networks, and disaster services on one agenda. The next step is not another isolated platform. It is a shared format through which 66 centers and 222 frontline social workers can collaborate without erasing local difference. Technology may organize and remind; relationships, cultural interpretation, professional judgment, and rights decisions must remain with people.

Sources

• Council of Indigenous Peoples, Taiwan: second 2026 in-service training session for Indigenous Family Service Center social workers
• Council of Indigenous Peoples, Taiwan: 2027 implementation plan for Indigenous Family Service Centers
• Ministry of Health and Welfare: Social Safety Net 2.0
• Laws and Regulations Database: Personal Data Protection Act
• Laws and Regulations Database: Disaster Prevention and Protection Act
• Laws and Regulations Database: Social Worker Act