Skip to main content

Brazilian AI Star Rio 3.5 Faces Backlash Over Alleged Model Plagiarism

Brazilian AI Project Rocked by Plagiarism Claims

What was supposed to be a triumph for Brazilian artificial intelligence has turned into controversy overnight. The Rio 3.5 model, developed by a municipal government-backed IT company in Rio de Janeiro, initially impressed the tech world with claims of cutting-edge performance. But less than 24 hours after its release, serious questions emerged about its true origins.

The Accusations Surface

Open-source AI group Nex-AGI dropped a bombshell analysis suggesting Rio 3.5 wasn't the original creation it claimed to be. Their forensic examination of the model's mathematical structure revealed what appears to be a patchwork of existing technologies.

"When we looked under the hood," explained a Nex-AGI spokesperson, "we found clear evidence that about 60% of Rio matches our Nex N2 Pro model, while the remaining 40% aligns with Alibaba's Qwen 3.5."

Smoking Gun Evidence

The research team provided compelling proof supporting their claims:

  • Identity Crisis: Removing Rio's system prompts caused the model to frequently identify itself as "Nex from Nex-AGI" in 79% of test cases, even repeating Nex's backstory verbatim
  • Mathematical Fingerprints: Statistical analysis showed precise 60/40 splits across all network layers - a pattern virtually impossible in genuinely original models

"This isn't just borrowing some ideas," the spokesperson added. "They've taken entire architectures and stitched them together."

Ethical Concerns in Open Source

The incident highlights growing tensions in the open-source AI community. While sharing and building on others' work is encouraged, the line between collaboration and plagiarism remains hotly debated.

Nex-AGI emphasized they support responsible open-source use, but stress that proper credit remains essential. "Our technology is meant to be built upon," they noted, "not repackaged and presented as someone else's breakthrough."

Key Points

  • Rio 3.5 claimed top-tier performance but faces serious plagiarism allegations
  • Independent analysis suggests it combines code from Nex N2 Pro (60%) and Alibaba's Qwen 3.5 (40%)
  • Technical evidence includes matching mathematical structures and unexpected identity behaviors
  • The case raises important questions about ethics in open-source AI development
  • Nex-AGI maintains they support collaboration but insist on proper attribution

As the AI field continues its rapid expansion, this incident serves as a cautionary tale about the importance of transparency and intellectual honesty in technological innovation.