> ## Content Index
> Fetch the complete content index at: https://www.ai-360.online/llms.txt
> Use this file to discover other available public pages before exploring further.

# Synapxe and AIDX partner on healthcare AI safety testing for HealthHub
- URL: https://www.ai-360.online/synapxe-aidx-healthhub-genai-safety-testing/
- Published: 2025-06-16T12:00:00.000Z
- Updated: 2026-10-26T17:20:32.000Z
- Description: Singapore HealthTech agency Synapxe signed an MoU with AIDX TECH to test AI model safety, including red-teaming of the HealthHub GenAI assistant.
- Author: AI-360
- Tags: Enterprise AI, news, #company-aidx, partnership

Synapxe, Singapore's national HealthTech agency, said it signed a memorandum of understanding with [AIDX](https://www.aidxtech.com/?ref=ai-360.online) TECH to identify AI risks in healthcare and use the AIDX platform to test AI models for safety and reliability, [according to a Synapxe media release dated 16 June 2025](https://www.synapxe.sg/media-releases/artificial-intelligence/ai-accelerate-2025?ref=ai-360.online).

Synapxe described collaboration elements including joint innovation labs, best-practice frameworks and specialised training, aligned with local and global frameworks for patient safety and public trust. An [AI Verify Foundation case study](https://assurance.aiverifyfoundation.sg/wp-content/uploads/2025/05/AIDX-x-Synapxe.pdf?ref=ai-360.online) details how AIDX tested Synapxe's RAG-based HealthHub GenAI conversational assistant, which serves health information from HealthHub website content over a multi-language speech path. The documented pipeline includes speech-to-text, PII detection and masking, content filtering, relevancy checks, knowledge-base search, a closed-source LLM response and text-to-speech, with chat history held in a secure cloud database.

AIDX's approach in the case study used proprietary benchmarking and adversarial red-teaming across robustness, ethics, privacy, toxicity and security — about 500 benchmark cases, 50 seed plus roughly 700 synthetic adversarial cases, and 14 named attack methods, with agent-based multi-turn red teaming up to 10 iterations and a consolidated Safety Score. AIDX is also listed among [AI Verify Foundation accredited testers](https://aiverifyfoundation.sg/ai-assurance/accredited-testers/?ref=ai-360.online) for benchmarking hallucination, undesirable content, data leakage, adversarial prompts and bias and fairness; accreditation recognises testing competencies and does not certify tested systems as safe. MoU scope and case-study metrics are Synapxe and AIDX statements; AI-360 has not independently re-run the tests.