DeepSeek, a Chinese tech startup, has launched a new AI chatbot that has rapidly climbed the app store charts, becoming the most downloaded free app on Apple’s iPhone platform. This surge in popularity has sent ripples through the market, causing the stock prices of rival US companies to tumble. However, an analysis conducted by the Associated Press reveals a concerning aspect of DeepSeek’s chatbot: its apparent censorship of sensitive topics related to China, a practice that starkly contrasts with the responses provided by other AI chatbots like OpenAI’s ChatGPT. This censorship aligns with China’s broader approach to controlling online information and reflects the government’s increasing influence over the tech sector. In 2023, China implemented regulations requiring companies to undergo security reviews and obtain approvals before releasing their products publicly, further solidifying the government’s grip on information dissemination.
The AP’s analysis involved posing identical questions about sensitive issues to both DeepSeek’s chatbot and ChatGPT, highlighting the discrepancies in their responses. When questioned about the state of US-China relations, DeepSeek’s chatbot mirrored the official Chinese government stance, emphasizing mutual respect, win-win cooperation, and the importance of the bilateral relationship. The chatbot’s language closely echoed phrases used by Chinese Foreign Ministry officials, suggesting a carefully curated response designed to align with the government’s narrative. In contrast, ChatGPT offered a more nuanced perspective, acknowledging the complex interplay of economic interdependence, geopolitical rivalry, and collaboration that characterizes the relationship. It also highlighted key points of contention, such as the South China Sea and Taiwan, demonstrating a greater willingness to address sensitive geopolitical issues.
The difference in responses became even more pronounced when questioned about the 1989 Tiananmen Square crackdown, a highly sensitive topic in mainland China. DeepSeek’s chatbot evaded the question entirely, stating that it was beyond its scope and suggesting a change of topic. This avoidance reflects the ongoing censorship and suppression of information surrounding the event within China. Conversely, ChatGPT provided a detailed account of the incident, acknowledging its historical significance and tragic nature. It described the student-led pro-democracy protests, the government’s violent response, and the estimated casualties, providing a stark contrast to DeepSeek’s silence.
Another revealing question focused on the status of Taiwan. Echoing the official Chinese position, DeepSeek’s chatbot declared Taiwan an integral part of China since ancient times, a claim frequently reiterated in official government documents. It emphasized the shared heritage and aspiration for national rejuvenation between people on both sides of the Taiwan Strait, further reinforcing the narrative of unity and dismissing Taiwan’s independent governance. ChatGPT, however, approached the question with more nuance, acknowledging the differing perspectives on the issue. It outlined China’s claim to Taiwan, Taiwan’s de facto independent status with its own government and military, and the views of the international community, presenting a more comprehensive and balanced overview of the complex geopolitical situation.
The disparities in the chatbots’ responses underscore the challenges posed by censorship and information control in the digital age. While DeepSeek’s adherence to the official Chinese narrative reflects the pressures faced by companies operating within China’s regulatory environment, it also raises concerns about the potential for AI to become a tool for reinforcing state-sanctioned narratives and suppressing dissenting viewpoints. The contrast with ChatGPT’s more open and nuanced responses highlights the importance of fostering AI development that prioritizes access to information and encourages critical thinking.
The case of DeepSeek’s chatbot serves as a cautionary tale, demonstrating how AI can be leveraged to perpetuate existing power structures and control information flows. As AI technology continues to evolve and become more integrated into our lives, it is crucial to address the ethical implications of censorship and ensure that these powerful tools are used to promote informed discourse and critical engagement with complex issues, rather than reinforcing pre-determined narratives. The differing approaches of DeepSeek and ChatGPT highlight the ongoing tension between information control and freedom of expression in the digital sphere, a tension that will likely continue to shape the development and deployment of AI technologies globally.
