UURL 財經新聞FINANCE
UURL Finance · Taiwan market + in-depth rewrites · hourly

91APP Hit by Chinese AI Tool Attack, 5 E-commerce Firms Affected

Browser TTS · no download

30-second brief

The Legislative Yuan Transportation Committee reviewed the MODA budget. Minister Lin Yi-jing confirmed that 91APP and about 5 to 6 e-commerce firms were scanned and penetrated by the Chinese Agentic AI tool ARTEX, mainly due to two open-source software programs missing patch updates; the CEC case was a vendor coding flaw, now patched, with 32 downloading IP addresses under investigation.

CEC Case Was Coding Flaw, 32 IP Addresses Under Investigation

The Legislative Yuan Transportation Committee reviewed the budget proposals for the Ministry of Digital Affairs and the National Communications Commission, with Minister Lin Yi-jing leading officials to report and answer questions. Regarding the leak of sensitive personal data including ID numbers of multiple mayoral and councilor candidates from the Central Election Commission's public official candidate property declaration system for 11 days, Lin clarified that from a technical perspective it was not an external hacker attack, but an avoidable error by the vendor when writing the web program, a flaw in the project software itself.

Administration for Cyber Security Director-General Tsai Fu-long added that after the incident a third-party forensics unit is making every effort to trace the sources of 32 downloading IP addresses, and the Administration for Cyber Security has introduced AI source-code detection tools to help the CEC inspect and strengthen system vulnerabilities. Lin promised to aim at establishing Uni-President unified defense and detection standards to prevent administrative agencies' personal data from being exposed to risks again.

91APP Penetrated by ARTEX, About 5 to 6 E-commerce Firms Affected

Regarding the suspected personal data leak from attacks on 91APP and other e-commerce platforms, Lin explained that according to the administrative investigation, the main cause was that the companies internally used two open-source software programs without timely installing the latest security patches. The investigation also confirmed that 91APP and the victimized e-commerce firms were indiscriminately scanned and penetrated by ARTEX, an Agentic AI automated attack tool developed in China. The same tool recently also victimized multiple financial institutions in Japan and South Korea.

Currently the Ministry of Digital Affairs has identified and is investigating about 5 to 6 key affected e-commerce firms, and requires companies to update patches immediately upon receiving cybersecurity alerts. Regarding the Bank SinoPac illegal transfer incident, Lin said the case involved software from a major foreign vendor, possibly due to unpatched known vulnerabilities or new attack methods. The Administration for Cyber Security has dispatched personnel to assist the Financial Supervisory Commission in the investigation, and will subsequently use AI tools to proactively conduct security testing for financial institutions.

Satellite Resilience Shifts to Leo, Strengthening Deepfake and Geospatial Data Protection

On communications resilience, Lin revealed that because Starlink assessed Taiwan's 4G/5G coverage at 99.9% and has low willingness to land in Taiwan, the Ministry of Digital Affairs has shifted to cooperating with Amazon Leo low-orbit satellites, and completed high-latitude field tests in Australia. Of the planned 2,200 terminal sites, 40% will be allocated to local governments for extreme-disaster rescue needs.

In countering AI disinformation, regarding the case of Taipei mayoral candidate Shen Boyang being smeared by AI deepfake audio, after the case was transferred the Ministry of Digital Affairs completed cross-platform takedowns in only 73 minutes, but front-end police forensics and reporting procedures took 6 days, and the Ministry is coordinating to establish encrypted and expedited forensic mechanisms. Regarding foreign disinformation with Facebook accounting for 70% and a doubled share on Threads, it will implement three principles of proactive disclosure by uploaders, technical labeling by platforms, and referral for prosecution for non-disclosure. The National Institute of Cyber Security has obtained U.S. government approval to cooperate with four major AI companies Anthropic, OpenAI, Google, and Microsoft on automated vulnerability scanning. Household registration and health insurance systems will introduce AI detection tools starting in December, and high-precision geographic map controls and AI Basic Act risk guidelines are under study.

FAQ

Why were 91APP and other e-commerce firms breached?

The Ministry of Digital Affairs' administrative investigation指出indicated the main cause was companies using two open-source software programs without timely security patch updates, and being indiscriminately scanned and penetrated by ARTEX, an Agentic AI automated attack tool developed in China. About 5 to 6 key e-commerce firms are currently known to be affected.

Was the CEC candidate data leak a hacker attack?

Lin Yi-jing clarified it was not an external hacker attack, but negligence by the vendor in writing the web program, a flaw in the project software itself. The Administration for Cyber Security has helped patch the vulnerability, and a third-party forensics unit is tracing the sources of 32 downloading IP addresses.

How will the Ministry of Digital Affairs strengthen AI-era cybersecurity?

The National Institute of Cyber Security is cooperating with Anthropic, OpenAI, Google, and Microsoft on automated vulnerability scanning, will introduce in-depth AI testing of core source code for household registration and health insurance systems starting in December, and promote unified defense standards and resident cybersecurity mechanisms.