Huawei’s AI Scandal: Was Its New Pangu Model Secretly Copied From Rival Alibaba?

The AI Plot Twist: Is Huawei’s Pangu a Secret Twin of Alibaba’s Qwen?

An anonymous GitHub post, a shocking allegation of "extraordinary correlation," and a tech giant's vehement denial. The world of artificial intelligence has been rocked by a scandal that reads like a tech-thriller. At the center of it all are two of China’s biggest tech titans: Huawei and Alibaba.

A group calling itself HonestAGI dropped a bombshell analysis on GitHub, claiming that Huawei’s new Pangu Pro AI model is suspiciously similar to Alibaba’s Qwen 2.5. This wasn't just a casual observation; the group alleged a level of technical overlap that suggested Huawei might have "upcycled" or copied its rival's work. The accusation sent shockwaves through China's AI community, raising uncomfortable questions about innovation, ethics, and the intense pressure to succeed in the global AI race.

Huawei’s research division, the renowned Noah's Ark Lab, was quick to fire back. They staunchly denied the claims, asserting that Pangu was an original creation, built from the ground up on their own proprietary Ascend AI chips. But the plot thickened when a self-proclaimed Huawei insider emerged, adding fuel to the fire with claims of systemic model cloning driven by a desperate need to keep pace with competitors. The original GitHub post has since vanished, but the controversy it ignited is far from over.

This incident peels back the curtain on the fierce, often unseen, rivalries brewing within China's AI sector. What was once seen as a more collaborative national effort now appears to be a hotbed of tension and distrust. If a major player like Huawei can be credibly accused of "re-skinning" a competitor's model, what does that mean for the future of open-source AI in China and the nation's standing in the global tech community? Let's unpack this complex and rapidly unfolding story.

A Tale of Two Models: Pangu vs. Qwen

Before we get into the nitty-gritty of the accusations, it's helpful to understand the two AI models at the heart of this controversy. While both are large language models (LLMs), they serve different purposes and target distinct markets.

This table highlights a key strategic divergence. Huawei, with its deep roots in enterprise hardware and government contracts, has tailored Pangu for industrial applications. Think of it as the AI workhorse for critical sectors. Alibaba, on the other hand, is aiming its Qwen models at the broader consumer market, creating tools that are more akin to ChatGPT. This difference in strategy is what makes the copying allegations so surprising—and so serious.

The Accusation That Shook the Industry

The controversy erupted when a previously unknown group, HonestAGI, published a technical paper on the code-sharing platform GitHub. The paper presented a detailed analysis claiming an "extraordinary correlation" between Huawei's Pangu Pro MoE (Mixture of Experts) model and Alibaba's Qwen 2.5-14B. The term "Mixture of Experts" refers to a sophisticated AI architecture that uses specialized sub-models to handle different tasks, making the overall model more efficient.

HonestAGI’s report suggested that the similarities were too significant to be coincidental. The group alleged that Huawei had engaged in "upcycling"—taking a pre-existing model and building upon it without starting from scratch. This, they argued, could point to several serious issues:

• 📌 Copyright Infringement: If Huawei used Qwen's architecture or code without the proper license or attribution, it could be a violation of intellectual property rights.
• 📌 Fabricated Technical Reports: The report implied that Huawei may have misrepresented its own development efforts and the resources invested in training Pangu.
• 📌 Erosion of Trust: Such practices, if true, could undermine the credibility of China's burgeoning open-source AI movement.

The technical claims, including a reported "correlation coefficient of 0.927," were specific enough to be taken seriously by many in the AI community, sparking intense online discussions.

Huawei's Staunch Defense and the Mysterious Whistleblower

Faced with a burgeoning scandal, Huawei's Noah's Ark Lab issued a swift and firm denial. The lab, which is responsible for Huawei’s AI research, released a statement asserting that:

• ✅ Pangu Pro was independently developed. The company insisted that the model was not based on "incremental training of other manufacturers' models."
• ✅ It was built on proprietary hardware. Huawei emphasized that Pangu is the first large-scale model built entirely on its own Ascend AI chips, a point of national pride and a key part of its tech self-reliance strategy.
• ✅ It featured key innovations. The lab stated that its team had made significant breakthroughs in architectural design and technical features.
• ✅ Open-source licenses were respected. While acknowledging the use of some third-party open-source code, as is common in AI development, Huawei maintained that it had strictly adhered to all licensing requirements.

However, just as Huawei was trying to control the narrative, another layer of complexity was added. A new GitHub post, this time from a self-proclaimed Huawei employee on the Pangu team, appeared to corroborate the original allegations. This anonymous insider claimed that the copying was not an isolated incident but a systematic practice born out of immense internal pressure to catch up with rivals like Alibaba and the rapidly rising AI startup DeepSeek.

The alleged whistleblower’s post painted a picture of a "delivery factory" culture where corners were cut to meet aggressive deadlines. It specifically claimed that one team had "cloned" Alibaba's Qwen-1.5 model, wrapped it in additional layers, and sold it to clients as a new version of Pangu. It even alleged that another model from the startup DeepSeek had been similarly copied. These claims, though unverified, added a significant amount of fuel to the fire and deepened the sense of crisis.

The Bigger Picture: Cracks in China's AI Ascent

This incident is more than just a corporate spat; it reveals the intense pressures and growing pains within China's AI industry. For years, the narrative has been one of a unified national push to challenge the dominance of Western tech giants like Google and OpenAI. However, this controversy suggests a different reality, one defined by fierce domestic competition.

The AI race in China has become a "brutal competition," with startups and tech giants alike trying to one-up each other. This has led to aggressive price wars and an almost frantic pace of model releases. In such an environment, the temptation to take shortcuts can be immense.

The Pangu-Qwen affair raises several critical questions:

• 🤔 What constitutes originality in AI? With many models using similar architectures and being trained on overlapping datasets, proving novelty is a significant challenge. This controversy highlights the need for clearer standards and verification methods.
• 🤔 Can open-source collaboration survive in a hyper-competitive market? The open-source movement relies on trust and good faith. Incidents like this could make companies more hesitant to share their work, potentially slowing down innovation.
• 🤔 How will this affect global trust in Chinese AI? As Chinese tech companies expand into international markets, particularly in Southeast Asia and the Middle East, their reputation for originality and ethical practices will be crucial. This scandal could give potential partners and customers pause.

Where Do We Go From Here? A Test of Credibility

The dust from this controversy has yet to settle. Alibaba has remained notably silent on the issue, which some analysts interpret as a strategic move to avoid escalating the conflict publicly while potentially exploring other options. The identity of HonestAGI and the alleged Huawei whistleblower remains unknown, adding a layer of mystery to the whole affair.

Some critics have pointed out potential flaws in HonestAGI's original analysis, including claims of fake references, which complicates the technical argument. Huawei, for its part, has reportedly offered to conduct an independent audit of Pangu's source code to prove its originality.

What is clear is that this is a pivotal moment for China's AI ecosystem. The outcome could set a precedent for how intellectual property disputes are handled and how the industry polices itself. As one analyst put it, in the AI race, "trust is the ultimate currency." For Huawei, and indeed for the entire Chinese tech sector, rebuilding that trust is now the primary challenge. The world will be watching to see if they can prove that their impressive technological advancements are not just borrowed from their rivals, but genuinely their own.

For those interested in exploring Alibaba's AI work further, you can find more details about their models on the Alibaba Cloud website.

A New Chapter in the AI Arms Race

The Huawei-Alibaba AI controversy is a stark reminder that the journey to artificial general intelligence is not just about technical breakthroughs. It's also about ethics, trust, and the very human pressures of competition. The allegations of model copying have pulled back the curtain on the intense rivalries that are shaping the future of AI in China.

This episode will likely force a period of introspection within the industry, prompting a much-needed conversation about transparency, attribution, and the rules of engagement in an open-source world. Whether this leads to stronger ethical guidelines and greater accountability or a more closed-off, distrustful ecosystem remains to be seen. One thing is certain: the race for AI supremacy just got a lot more complicated.