Controversy Over IplanRIO AI Model in Rio de Janeiro
The IT company IplanRIO, under the Rio de Janeiro city government, issued a statement responding to the model controversy, admitting that the Rio-3.5-Open-397B model is a fine-tuned combination of the Nex-N2 Pro and Alibaba's Qwen 3.5 open-source models, and is not a self-developed base model.
The official statement mentioned that the final weight file was lost, leading to the upload of an intermediate comparison version, and apologized for not acknowledging the Nex-AGI team. The Hugging Face page has been taken down and will need to be retrained before being released.
The Rio city government previously allocated about $100,000 for training expenses, raising strong suspicions of misappropriation of public funds.
Source: Public information
ABAB AI Insight
IplanRIO, as a municipal IT company, has exposed execution and oversight gaps in local government digital transformation through this AI project controversy. The "lost file" defense further amplifies public distrust in the efficiency of public fund usage.
On the capital front, local government AI budgets may face stricter audits, shifting funding from concept-driven projects to suppliers with actual delivery capabilities, motivated by the desire to avoid similar scandals and enhance transparency in public projects.
Similar controversies arising from ineffective execution of AI bidding projects by various local governments, along with historical cases of corruption in public procurement, indicate that emerging market local governments are in the early stages of transitioning from enthusiastic investment to compliance oversight in digitalization. This incident may accelerate the normalization of procurement processes.
Essentially, this reflects regulatory changes, where deviations in the execution of public AI projects lead to accountability. The mechanism is that the accessibility of open-source models lowers technical barriers, but the lack of oversight can lead to waste of funds, concentrating capital and public attention on transparent and verifiable projects.
ABAB News · Cognitive Law
Open-source is easy to shell, but public funds are hard to fool; transparency is the best firewall.
Concept hype is fast, but actual delivery is slow; government projects rely on oversight rather than enthusiasm.
Short-term scandal fermentation, mid-term tightening of procurement, and long-term evolution of public AI towards specialization.