Open-source software underpins 98% of commercial applications and 100% of the 500 most powerful supercomputers in the world, yet its maintenance relies heavily on community expertise. At Software Freedom Day 2026 hosted by the Vietnam Free and Open Source Software Club (VFOSSA) in collaboration with the Faculty of Information Technology and Digital Economy at the Banking Academy, technology leaders highlighted critical strategies for mastering artificial intelligence, securing open-source supply chains, and navigating strict compliance mandates.
The Economic Reality of Open-Source Maintenance
Global open-source development moves billions of dollars, but the true asset isn’t just capital. Citing the Open Source Grants Report 2024 from Harvard University, the Linux Foundation, and GitHub, Dr. Nguyen Hong Quang, former chairman of VFOSSA, noted that the global ecosystem receives roughly $7.7 billion annually. However, 86% of that value stems directly from the expertise of professionals.
Time spent on maintenance, bug fixing, and community support keeps the digital world running. Without active intervention, projects fall victim to the “tragedy of the commons.” When users lack responsibility for the further development and maintenance of software, the supply chain fractures. Modern software engineering demands shared responsibility. Enterprises and students alike must step up through reporting errors, writing documentation, and guiding new users.
AI-Generated Code and the Responsibility Crisis
Artificial intelligence writes code at unprecedented velocities. That speed introduces questions regarding software license compliance—spanning MIT, Apache 2.0, GPL, and LGPL—alongside security and accuracy.
Google AI trainer and speaker Tran Bui Xuan addressed these exact pressures during the recent tech gathering in Vietnam. The philosophy moving forward is simple: responsibility before speed. Engineering teams cannot blindly deploy LLM-generated snippets straight to production. Instead, a “human-in-the-loop” model is mandatory. AI should assist with explaining the codebase, generating test cases, or restructuring the code. Automated verification tools like OSV-Scanner must run alongside security tests and approval by the maintainer before any pull request merges.
True mastery of artificial intelligence requires three distinct tiers:
- The ability to use it for acceleration.
- The ability to implement it to develop functionality.
- The ability to control it to build long-term trust.
Closing the Gap in Enterprise AI Adoption
The rise of open AI architectures, including models like DeepSeek and Llama, has fundamentally shifted enterprise software strategies. The quality gap between closed AI and open AI has shrunk to 3.3%.
According to Dr. Dang Minh Tuan, Vice President of CMC Technology Group, this convergence opens doors for the banking industry. Financial institutions can now deploy internal Retrieval-Augmented Generation (RAG) solutions. By doing so, banks can meet regulatory frameworks such as Law 91/2025/QH15 on the protection of personal data and Law 134/2025/QH15 on artificial intelligence.
