Venezuela could soon be among the first states, apart from China, to adopt the new wave of Chinese artificial intelligence systems designed to monitor the population and exercise political control. That is the claim of the Australian Strategic Policy Institute (ASPI). Its study, funded by the Human Rights Foundation, was published in September 2026. ProBox published an article about it on 21 September 2026, and Global Voices later circulated an edited version.
The pact with iFlytek
According to the study, in July 2025 Caracas signed a memorandum of understanding with iFlytek, a Chinese company that has been on the US trade sanctions list since 2019 for its involvement in the repression of Muslim minorities in Xinjiang. The agreement provides for developing artificial intelligence jointly on Venezuelan soil. It was negotiated by Delcy Rodríguez, vice president at the time and now interim president.
The text has not been made fully public, as has been the case for years with other technology agreements between Venezuela and China. That opacity, the report notes, makes it impossible to gauge the real scope of the pact, which is why its authors call for these agreements to be published in full.
A surveillance base already in place
The researchers say the country is not starting from scratch: for 20 years it has been buying identification equipment, cameras, command centres and networks from Chinese manufacturers. In this web they identify six interlinked technology layers: the telecommunications network, video surveillance, internet filtering, unified data records, digital identification and, as the most recent addition, AI-assisted analysis. Huawei, ZTE, CEIEC, Hikvision and Dahua are among the firms that helped build or strengthen them.
The centrepiece of digital identity is the Carnet de la Patria, tied to social benefits. According to the report, in January 2023 it had 21.8 million registered users, roughly three quarters of the population, with a ceiling of 14.2 million people served each month. Video surveillance relies on command centres such as VEN911 and SIMA. Although the whole is fragmented, uneven and subject to technical limits, it gives state institutions the ability to recognise and track the population, and to intervene based on their behaviour.
What AI would add
For ASPI, iFlytek's technology could assemble these pieces into a larger structure, speed up and extend surveillance, and make it harder for anyone to escape it. The company, one of the world's leading providers of speech recognition and content moderation, offers, according to the study, a tool capable of recognising "more than a hundred types" of sensitive political material in voice recordings. It also sells voiceprint identification, the same technology that, according to ASPI, it has supplied to public security forces in Xinjiang.
If that capability were combined with the cameras already deployed and the data stored by the Carnet de la Patria, the authors say, the government could flag someone pre-emptively because of their associations, movements or social media activity, even before any specific act.
The authorities themselves acknowledged a precedent. In January 2025 a command centre opened that brings together data from thousands of communal councils and circuits: the Sala Nacional de Autogobierno Popular (National Room of Popular Self-Government). Nicolás Maduro said at the time that it operated "with Chinese technology and artificial intelligence."
Political transition and recommendations
The pact was sealed months before US forces captured Maduro in January 2026. Rodríguez now runs the country with Washington's support, but the report argues that the surveillance apparatus has not been dismantled under her leadership. According to the text, the UN fact-finding mission documented that the infrastructure continued operating after the handover.
The authors see an opportunity in the interim government's dependence on the United States, which could be used to demand the annulment of the memorandum before the technology is integrated. The report puts forward seven recommendations, including:
- financial support, protection and cybersecurity training for journalists and groups under threat, as well as faster humanitarian visas;
- free access to AI tools to circumvent censorship;
- that the United States ask Rodríguez to cancel the agreement and deactivate mechanisms such as the Carnet de la Patria;
- drafting of privacy and oversight laws for a future political opening;
- sanctions on officials and companies that contribute to digital repression;
- security criteria for foreign AI used by governments, with independent audits and clarity on the origin of the models.
The work is authored by Fergus Ryan, Bethany Allen and Nathan Attrill of ASPI's Cyber, Technology and Security Program. Mariví Marín of ProBox, Shannon van Sant and Alex Colville contributed.
Sources
- Global Voices en Español (CC BY 3.0)
