The Verge56%

China delivers a one-two punch to America’s AI dominance 74%

By Robert Hart53%

7/20/2026, 3:16:33 AM

BS Summary: This article contains 28 faulty reasoning types, including Hasty Generalization, Unattributed Quote, and False Dilemma, with Biased Writer Voice as the most egregious example at 45.2% saturation with 260 hits. Analysis detected 1,840 faulty-reasoning hits from 575 analyzed words, generating a BS Score of 66.1% and a BS Rank of 74% (5,735 of 21,887 articles). This article is worse (more manipulative) than 73.80% of the article peer group.

China’s leading AI companies are ramping up the pressure on Silicon Valley, as Moonshot and Alibaba unveiled models they claim can go toe-to-toe with the best from OpenAI and Anthropic at a fraction of the cost. 
The rapid-fire releases suggest America’s lead at the AI frontier is increasingly tight, just as the technology is becoming central to national security, economic power, and geopolitical influence. 
The opening salvo came from Beijing-based Moonshot AI, one of China’s leading AI model developers, which unveiled Kimi K3 on Friday. 
Moonshot claims its own testing ranks it consistently above nearly every US system, trailing only OpenAI’s GPT-5.6 Sol and Anthropic’s Claude Fable 5, though it came out ahead on certain benchmarks. 
Over the weekend, Chinese tech behemoth Alibaba followed with a preview of Qwen3.8, a new model it says is “one of the most powerful model[s] available today” and “second only to Fable 5,” Anthropic’s flagship model. 
Both companies are emphasizing a key difference from the leading US labs: Rather than locking their most advanced models behind closed doors, they are making them publicly available. 
While some US companies, most notably Meta, have taken a similar approach, releasing models that developers can freely download, modify, and build upon has become a growing point of differentiation for China’s AI industry. 
Moonshot describes Kimi K3 as the world’s largest open-source AI system, with 2.8 trillion parameters. 
Parameter counts are measures of a model’s complexity during training and offer a rough indication of its scale and performance, though bigger does not always mean better. 
Alibaba says Qwen3.8 is a 2.4 trillion parameter model and “continuously evolving.” 
Neither OpenAI nor Anthropic disclose exact parameter counts for their leading systems. 
It remains difficult to assess how capable either Chinese model is until they are fully released and independently tested. 
Moonshot says it will release full model weights  the internal numerical learned during an AI model’s training period  a week from now on July 27th. 
Alibaba says Qwen3.8 is “going open-weight soon.” 
Even so, the releases have already sharpened competition between the US and China in what has been repeatedly characterized as the defining technological race of our time. 
They have shaken up the industry in a way not seen since DeepSeek unveiled a low-cost model last year that rivaled leading US systems. 
The models also raise questions about whether the vast sums of money US companies are pouring into chips, data centers, and model training can secure a durable advantage, particularly if Chinese rivals can approach  or surpass  that frontier with fewer resources. 
The prospect of two highly capable Chinese models being released for others to download and adapt also contrasts starkly with the more guarded approach of US labs, whose most advanced systems remain proprietary. 
That openness emerges even as Washington moves rapidly to restrict global access to the underlying technology. 
The government has used export controls to restrict China’s access to the most advanced chips, as well as to force Anthropic to pull its most capable system from the market over concerns it could help foreign competitors catch up. 
Whether the new Chinese models live up to their creator’s claims remains to be seen. 
But, like DeepSeek before them, they are likely to sharpen the technological rivalry between the US and China, influence economic and national security policy, and show that America’s lead is far narrower than it once appeared. 
Article reasoning-pattern comparisonThis article: 6.3%Robert Hart: 4.3%The Verge: 3.6%Confirmation Bias6.3%This article: 0.0%Robert Hart: 1.9%The Verge: 1.3%Anchoring Bias0.0%This article: 10.4%Robert Hart: 5.8%The Verge: 3.8%Availability Heuristic10.4%This article: 5.9%Robert Hart: 0.8%The Verge: 1.3%Representativeness Heuristic5.9%This article: 0.0%Robert Hart: 2.4%The Verge: 0.7%Hindsight Bias0.0%This article: 3.3%Robert Hart: 1.4%The Verge: 1.7%Overconfidence Bias3.3%This article: 13.7%Robert Hart: 10.2%The Verge: 6.2%Framing Effect13.7%This article: 0.0%Robert Hart: 1.1%The Verge: 0.9%Loss Aversion0.0%This article: 4.9%Robert Hart: 0.1%The Verge: 0.5%Status Quo Bias4.9%This article: 0.0%Robert Hart: 1.3%The Verge: 0.4%Sunk Cost Effect0.0%This article: 14.8%Robert Hart: 1.7%The Verge: 4.0%Optimism Bias14.8%This article: 12.3%Robert Hart: 3.2%The Verge: 2.4%Pessimism Bias12.3%This article: 7.5%Robert Hart: 6.9%The Verge: 9.7%Negativity Bias7.5%This article: 16.3%Robert Hart: 1.7%The Verge: 1.5%Self-Serving Bias16.3%This article: 0.0%Robert Hart: 0.0%The Verge: 0.8%Fundamental Attribution Error0.0%This article: 6.8%Robert Hart: 0.4%The Verge: 0.2%Actor-Observer Bias6.8%This article: 5.7%Robert Hart: 0.8%The Verge: 0.6%In-Group Bias5.7%This article: 0.0%Robert Hart: 0.8%The Verge: 0.4%Out-Group Homogeneity Bias0.0%This article: 6.3%Robert Hart: 1.0%The Verge: 2.8%Halo Effect6.3%This article: 0.0%Robert Hart: 0.0%The Verge: 0.2%Horn Effect0.0%This article: 0.0%Robert Hart: 0.0%The Verge: 0.0%Dunning-Kruger Effect0.0%This article: 15.8%Robert Hart: 5.0%The Verge: 1.8%Recency Bias15.8%This article: 0.0%Robert Hart: 0.4%The Verge: 0.4%Primacy Effect0.0%This article: 2.6%Robert Hart: 0.0%The Verge: 0.1%Blind-Spot Bias2.6%This article: 0.0%Robert Hart: 0.2%The Verge: 1.1%Ad Hominem0.0%This article: 0.0%Robert Hart: 0.0%The Verge: 0.3%Straw Man0.0%This article: 14.3%Robert Hart: 4.3%The Verge: 4.0%Appeal to Authority14.3%This article: 18.1%Robert Hart: 5.6%The Verge: 1.6%False Dilemma18.1%This article: 7.5%Robert Hart: 1.3%The Verge: 1.2%Slippery Slope7.5%This article: 0.0%Robert Hart: 0.0%The Verge: 0.2%Circular Reasoning0.0%This article: 22.6%Robert Hart: 8.0%The Verge: 6.8%Hasty Generalization22.6%This article: 0.0%Robert Hart: 0.0%The Verge: 0.2%Red Herring0.0%This article: 4.7%Robert Hart: 1.5%The Verge: 0.7%Bandwagon4.7%This article: 6.8%Robert Hart: 4.1%The Verge: 4.2%Appeal to Emotion6.8%This article: 0.0%Robert Hart: 1.9%The Verge: 1.1%Begging the Question0.0%This article: 16.5%Robert Hart: 2.0%The Verge: 2.3%Post Hoc (False Cause)16.5%This article: 0.0%Robert Hart: 0.0%The Verge: 0.2%Tu Quoque0.0%This article: 2.6%Robert Hart: 1.3%The Verge: 0.8%Burden of Proof2.6%This article: 0.0%Robert Hart: 0.0%The Verge: 0.3%Appeal to Nature0.0%This article: 0.0%Robert Hart: 0.4%The Verge: 0.3%Composition/Division0.0%This article: 0.0%Robert Hart: 2.4%The Verge: 3.6%Anecdotal0.0%This article: 0.0%Robert Hart: 0.0%The Verge: 0.1%No True Scotsman0.0%This article: 16.0%Robert Hart: 1.0%The Verge: 2.2%Ambiguity (Equivocation)16.0%This article: 0.0%Robert Hart: 0.0%The Verge: 0.0%Gambler’s Fallacy0.0%This article: 0.0%Robert Hart: 0.1%The Verge: 0.1%Middle Ground0.0%This article: 0.0%Robert Hart: 0.0%The Verge: 0.1%Personal Incredulity0.0%This article: 0.0%Robert Hart: 0.0%The Verge: 0.2%Special Pleading0.0%This article: 0.0%Robert Hart: 0.0%The Verge: 0.2%Genetic Fallacy0.0%This article: 22.3%Robert Hart: 3.9%The Verge: 2.6%Unattributed Quote22.3%This article: 0.0%Robert Hart: 1.6%The Verge: 1.4%Quote-first Misdirection0.0%This article: 45.2%Robert Hart: 2.3%The Verge: 10.8%Biased Writer Voice45.2%This article: 4.9%Robert Hart: 1.5%The Verge: 1.1%Indoctrination4.9%This article: 5.9%Robert Hart: 0.5%The Verge: 1.6%Politically Left Leaning Bias5.9%This article: 0.0%Robert Hart: 0.0%The Verge: 0.1%Politically Right Leaning Bias0.0%This article: 0.0%Robert Hart: 1.5%The Verge: 4.2%Attempt to Sell a Product or S…0.0%

575 words analyzed.

Speakers

2speakers22%attributed speech447writer words
Voice mapSelect a segment to jump to its words
Writer's voice • 9 words • 100.0% coverageWriter's voice • 36 words • 100.0% coverageWriter's voice • 28 words • 100.0% coverageWriter's voice • 21 words • 100.0% coverageMoonshot • 31 words • 100.0% coverageAlibaba • 36 words • 100.0% coverageWriter's voice • 28 words • 100.0% coverageWriter's voice • 34 words • 100.0% coverageMoonshot • 15 words • 100.0% coverageWriter's voice • 27 words • 0.0% coverageAlibaba • 12 words • 100.0% coverageWriter's voice • 12 words • 0.0% coverageWriter's voice • 19 words • 0.0% coverageMoonshot • 27 words • 100.0% coverageAlibaba • 7 words • 100.0% coverageWriter's voice • 27 words • 100.0% coverageWriter's voice • 24 words • 0.0% coverageWriter's voice • 43 words • 0.0% coverageWriter's voice • 33 words • 100.0% coverageWriter's voice • 16 words • 0.0% coverageWriter's voice • 39 words • 0.0% coverageWriter's voice • 15 words • 0.0% coverageWriter's voice • 36 words • 100.0% coverage
Selected voice

Alibaba

100%flagged-word coverage
55 attributed words43% of attributed speech97% writer coverage
0%50.0%100.0%Unattributed Quote+100.0 ptsWriter: 0.0%Alibaba: 100.0%100.0%Biased Writer Voice+15.3 ptsWriter: 50.1%Alibaba: 65.5%65.5%Politically Left Leaning B-7.6 ptsWriter: 7.6%Alibaba: 0.0%0.0%Indoctrination-6.3 ptsWriter: 6.3%Alibaba: 0.0%0.0%

Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.

Loading…
Loading…
Loading…

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.