The Financial Embedding of Generative AI: A Comprehensive Review of DeepSeek’s Innovation Dividends and Negative Externalities

Jialiang Liu

doi:10.6981/FEM.202608_7(8).0004

Authors

Jialiang Liu Swinburne, Shandong University of Science and Technology, Jinan, Shandong, 250000, China

DOI:

https://doi.org/10.6981/FEM.202608_7(8).0004

Keywords:

Generative Artificial Intelligence; Financial Embedding; Innovation Dividends; Negative Externalities; DeepSeek.

Abstract

Generative artificial intelligence is increasingly integrated into the financial sector. While offering substantial commercial benefits, it also introduces notable negative externalities, reflecting the dual nature of its application. Most prior studies focus solely on either the technological potential or the associated risks of generative AI in finance, lacking a comprehensive framework that simultaneously considers both innovation dividends and negative externalities. To address this gap, this study uses DeepSeek as a representative technological model, adopting a 'financial embedding' perspective to analyze how generative AI creates value and evolves risks within financial scenarios. By reviewing existing research, the paper reconciles fragmented findings from narrowly focused studies. Considering the technical features and practical deployment of domestic large language models, it clarifies the relationship between smart finance innovation and risk constraints, identifies value generated through technological support and business model upgrades, and examines operational challenges such as algorithmic errors, data security issues, and regulatory delays. Using a dual-in-one analytical approach, the study integrates prior research to propose a framework adapted to the local financial context, offering theoretical guidance and governance strategies for the sustainable and compliant integration of generative AI into financial systems.

Downloads

Download data is not yet available.

References

[1] Wang, Y. (2025). Cutting-edge advances in generative artificial intelligence in the financial sector. Oriental Forum: Journal of Qingdao University, Social Sciences Edition, (2), 43–53.

[2] Gibney, E. (2025). The Chinese finance whizz whose DeepSeek AI model stunned the world. Nature, 648(8094), 526. https://doi.org/10.1038/d41586-025-03845-4

[3] Liu, Y., & Wang, Y. (2025). Pathways and policy innovations for DeepSeek to empower China’s modernization: Seizing the opportunity of industrial restructuring. Journal of Zhengzhou University, Philosophy and Social Sciences Edition, 58(3), 68–76+144.

[4] Iovane, G. (2025). Sophimatics and 2D complex time to mitigate hallucinations in LLMs for digital transformation. Applied Sciences, 16(1), 288. https://doi.org/10.3390/app16010288

[5] Yitong, Z. (2026). Bias in large language models: Methods, evaluation, and prospects. ITM Web of Conferences, 84, 03007.

[6] Yang, D., & Huang, T. (2025). DeepSeek-style artificial intelligence empowering FinTech 3.0: Systemic transformation and regulatory adaptation. Journal of Xinjiang Normal University, Philosophy and Social Sciences Edition, 46(4), 80–89.

[7] Tao, Y. (2025). A study on AI-driven financial innovation models and regulatory challenges from the perspective of DeepSeek. Modern Business, (22), 147–150.

[8] Rahman, A., Mahir, H. S., Tashrif, A. T. M., et al. (2025). Comparative analysis based on DeepSeek, ChatGPT, and Google Gemini. Systems and Soft Computing, 7, 200396.

[9] Ge, C. (2026). The development and future applications of large language models. Technological Innovation and Science Management, 2(2).

[10] Li, Y., & Zhou, H. (2025). DeepSeek: Empowering inclusive finance and promoting common prosperity in rural areas. Contemporary Economic Management, 47(5), 86–96.

[11] Shi, Z., & Lu, M. (2025). DeepSeek driving the intelligent transformation of banks: Localised model optimisation and the leap in risk management. Regional Financial Research, (2), 1–9.

[12] Bu, Y. (2025). Legal risks and legal governance of DeepSeek-like artificial intelligence applications in the financial sector. Heilongjiang Finance, (11), 55–58.

[13] Wang, J. (2025). Building a future industrial ecosystem based on DeepSeek. Journal of Northwest University, Philosophy and Social Sciences Edition, 55(3), 71–80.

[14] Chen, X., & Zhu, X. (2025). Six dimensions of how DeepSeek technology breaks through to reshape the global AI industry competitive landscape: Paradigms and implications. Nanjing Social Sciences, (7), 41–57.

[15] Jiao, L. (2026). Practical exploration of DeepSeek empowering the intelligent transformation of counter services in bank digital transformation. Business Observer, 12(6), 78–81.

[16] Yang, Z. (2025). DeepSeek large models and the digital finance strategy for small and medium-sized banks. Northern Finance, (7), 21–25.

[17] Song, D. (2026). The value implications, internal logic and pathways of DeepSeek in empowering high-quality regional development. Science and Management [Online].

[18] Zeng, M. (2025). DeepSeek: Leading a technological leap into a new era of artificial intelligence and new prospects for finance. Chinese Financier, (4), 81–82.

[19] Didisheim, A., Fraschini, M., Somoza, L., et al. (2026). Is AI reasoning useful in finance? Finance Research Letters, 97, 109783.

[20] Yang, X. (2025). AI competition and firm value: Evidence from DeepSeek’s disruption. Finance Research Letters, 80, 107447.

[21] Li, S., & Xia, J. (2026). Legal dilemmas and response strategies for DeepSeek’s empowerment of consumer finance development. Journal of Shaanxi Normal University, Philosophy and Social Sciences Edition, 55(1), 179–192.

[22] An, L. (2026). Personal information security governance in the iteration of generative artificial intelligence: A perspective based on the evolution of large-scale model technology. Journal of Agricultural Library and Information Science, 38(4), 61–70.

[23] Su, J. (2025, February 25). DeepSeek: Opportunity or challenge? China Banking and Insurance News, Issue 008.

[24] Meng, T., & Lu, M. (2025). Exploring digital and intelligent transformation in tech-driven finance based on DeepSeek: Technological advantages, industry empowerment and future challenges. Enterprise Technology and Development, (3), 8–15.

[25] Li, H., Si, H., & Hai, M. (2025). Comparative analysis of domestic open-source large language models. Procedia Computer Science, 266, 955–962.