Nanjing's tax bureau recently presented the results of the first 10 priority tasks under its tax collection and administration upgrade and released a second batch of 12 tasks. The highlight is an AI smart review system for non-tax revenue. In the past, staff had to check more than 20 fields in each land-transfer contract by hand, spending over 15 minutes per case and risking oversight or misjudgement. Now the system extracts 37 key fields on its own, pulls public data for three-way comparison and completes pre-processing around the clock, so the front desk can simply use the result and finish an initial review in about three minutes. The system can identify 21 types of land-transfer risk, covering all risks listed in the relevant risk-control guideline, plus five types of sea-area usage fee risk, with accuracy above 98%. The bureau plans to extend the approach to seven other fee items filed per occurrence, including mineral resource revenue. It has also built digital profiles and one-file records for every taxpayer, tax-source analysis reports, and an assistant for the interfaces between different management duties, lifting the efficiency of cross-tax dispute verification in export business by nearly 50%.
News Detail
Nanjing Tax Bureau Advances Tax Administration Upgrade with AI Smart Review for Non-Tax Revenue Risk Control
Nanjing's tax bureau has showcased the first 10 priority tasks of its tax collection and administration upgrade and issued a second batch of 12 tasks. Using an AI large model, the bureau built a smart review system for non-tax revenue: it automatically extracts 37 key fields from land-transfer contracts, cross-checks public data and pre-processes cases around the clock, cutting initial review of one case from more than 15 minutes to about three minutes. The system now identifies 21 types of land-transfer risk and five types of sea-area usage fee risk with accuracy above 98%, and will be extended to seven other fee items. The bureau also introduced digital tools such as tax-source analysis reports and digital profiles of taxpayers.
2026-09-13