Maduro Is Gone, but His Surveillance State Has a Chinese AI Upgrade on Order

Casual observers of Venezuelan politics might have assumed that the U.S. capture and removal of former Venezuelan president Nicolas Maduro in January would guarantee the dismantling of his authoritarian regime and a rupture with Beijing.

Secretary of State Marco Rubio, reportedly in constant contact with Rodriguez via WhatsApp, said Washington had ‘tremendous control and leverage over what those interim authorities are doing and are able to do’.

Yet one of the most consequential pieces of Maduro’s authoritarian legacy remains intact.

As we show in our new report, Warning Signals: Venezuela and the risk of Chinese AI-enabled digital authoritarianism, Venezuela retains an apparatus of digital surveillance and political control built partly with Chinese technology. Before Maduro’s removal, Caracas had also begun preparing to add something new: Chinese artificial intelligence.

For two decades, Venezuela imported much of the plumbing of digital authoritarianism, such as telecommunications equipment, identity systems, cameras, command centres and technology for controlling the internet. In July 2025 it took a potentially more consequential step: signing an agreement with Chinese AI company iFlytek.

In January 2025, Maduro announced the creation of a national command room and ordered the establishment of 5,334 local data rooms, embedded in Venezuela’s communal governance system, to feed it. The national command room, he said, would operate with artificial intelligence and Chinese technology.

The largest component is the Carnet de la Patria, or Homeland Card, a digital identity and welfare system built with significant assistance from ZTE. Separately, Chinese companies including CEIEC, Huawei, Hikvision and Dahua have played varying roles in Venezuela’ s public-security cameras and command centres, its telecommunications networks and its internet-control infrastructure.

The system is not a seamless Chinese surveillance state transplanted to Latin America. It is fragmented, unevenly deployed and constrained by weak infrastructure. But AI could make these disparate systems far more useful to an authoritarian government by connecting citizen records, automating censorship, analysing camera feeds and combining data that today sits in separate databases. It could make repression cheaper, faster, more scalable and harder to see.

The company that Venezuela has chosen as its AI partner makes the risk particularly acute. iFlytek’s technology is embedded in China’s policing, judicial and censorship systems, and the company markets tools capable of detecting political and other sensitive content. The U.S. government placed it on an export-control blacklist in 2019 over its role in surveillance and repression of Muslim minorities in Xinjiang.

Yet in April 2025, Rodriguez visited iFlytek’s facilities in Beijing and declared that ‘only China’ could ensure countries in the Global South obtained AI ‘for development and not for domination’. Three months later, she signed the agreement covering the development of ‘sovereign’ AI technologies with iFlytek.

Rodriguez is central to this story. Before becoming acting president, she chaired the China–Venezuela High-Level Joint Commission. As executive vice president, she was also the highest civilian authority with formal oversight of Venezuela’s domestic intelligence agency. The person Washington is relying on to run post-Maduro Venezuela therefore helped oversee both the old apparatus of surveillance and the relationship that could provide its next technological upgrade.

That apparatus did not disappear with Maduro. In March 2026, UN investigators reported dozens of politically motivated detentions had followed his removal, while independent researchers have found Venezuela’s surveillance infrastructure remained ‘fully operational’. Rodriguez’s government has given no public indication that it intends to dismantle the system or abandon the iFlytek agreement.

The policy toolkit for countering digital authoritarianism was largely built for hardware, such as cameras, routers and telecommunications equipment, that can be identified, sanctioned, inspected or removed. AI changes that equation. What matters most may now be an intangible model whose training, filtering rules and assumptions about what constitutes a threat are largely invisible to the government importing it.

Venezuela offers an unusual opportunity to act before it is too late. The supplier is known, the agreement is signed and the infrastructure onto which AI could be layered already exists. But the AI surveillance capability itself is not yet publicly documented as operational. And Washington has extraordinary leverage over the government that signed the deal.

The U.S. should require Rodriguez to cancel the iFlytek agreement and use its leverage to begin dismantling, rather than merely inheriting, Maduro’s machinery of digital repression.

More broadly, democratic governments need standards for foreign AI used in sensitive state functions, including disclosure of where its training data came from, how it filters information, how updates are managed and who controls fine-tuning.

Venezuela is becoming a test of whether democracies can act against AI-enabled authoritarianism before the system is fully built. For once, the warning signs have arrived before the finished product.

Fergus Ryan is a China analyst with ASPI’s Cyber, Technology and Security program.

Image of (L-R) Delcy Rodriguez, Nicolas Maduro and Chinese ambassador Lan Hu at the July 2025 China–Venezuela Expo in Caracas, at which Venezuela and China signed a series of technology and cooperation agreements, including the iFlytek AI agreement: Prensa Presidencial de Venezuela.

AI contributed no ideas to this article – Fergus Ryan.

This article appeared originally at The Strategist (ASPI).