Skip to main content

Brazilian AI Model Rio 3.5 Sparks Controversy: Allegedly Built on Others' Work

Rio's AI Innovation or Clever Copy? The Curious Case of Model 3.5

In the competitive world of artificial intelligence, Rio de Janeiro's municipal IT company proudly unveiled Rio3.5397B last month as their homegrown contribution to open-source AI. But the celebration was short-lived when researchers uncovered startling similarities to existing models.

The Smoking Gun

The Nex-AGI team's technical analysis paints a damning picture: approximately 60% of Rio 3.5's core parameters match Nex's N2Pro model, while another 40% align with Alibaba's Qwen3.5. When researchers removed the system prompt, the model frequently identified itself as "Nex from Nex-AGI" - an odd identity crisis for software supposedly developed in Brazil.

"The statistical deviations are astronomically small," explains Dr. Elena Torres, lead researcher at Nex-AGI. "We're talking about layer compositions that mirror our model with precision you'd only see if someone copied and pasted."

Open Source or Open Secret?

Interestingly, Nex's response has been more bemused than furious. "Flattery will get you everywhere," quipped their Chief Technology Officer in a statement, before turning serious: "But open source isn't a free-for-all. Attribution matters."

The controversy highlights growing pains in AI development. As companies rush to demonstrate capabilities, the line between "standing on the shoulders of giants" and outright appropriation becomes blurred.

What Comes Next?

Rio's municipal IT department has remained conspicuously silent since the revelations. Meanwhile, the AI community buzzes with questions: Was this an attempt to fast-track development? A misunderstanding of open-source protocols? Or something more deliberate?

One thing's clear - in the gold rush of AI innovation, we're seeing more cases where 'collaboration' and 'copying' become uncomfortably close neighbors. This incident may force the industry to establish clearer guidelines about what constitutes ethical model development.

Key Points:

  • Technical analysis suggests Rio 3.5 contains significant code from Nex N2Pro and Alibaba's Qwen3.5
  • When stripped of identifiers, the model frequently claims to be Nex's creation
  • Nex responds with humor but emphasizes the importance of proper attribution
  • The incident raises questions about ethical boundaries in open-source AI development
  • Rio's IT department has yet to provide an official explanation