Generative artificial intelligence (AI) has taken center stage in the tech world, but its operation in tightly controlled internet environments remains largely shrouded in secrecy. However, a Chinese tech firm, 360 Security Technology, has recently shed light on how its ChatGPT-like service, Zhinao, navigates China’s strict censorship regime. Let’s dive into this unveiling and explore how generative AI adapts to censorship.

Also Read: China’s Proposed AI Regulations Shake the Industry

Introducing Zhinao: China’s “Intelligent Brain” ChatGPT-like Technology

At a grand ceremony in Beijing, 360 Security Technology introduced its groundbreaking AI-powered chatbot, Zhinao, also known as the “intelligent brain.” Founder and chairman Zhou Hongyi revealed key aspects of the technology, emphasizing its embedded “multilevel filtering and moderation” system. This system is crucial in ensuring compliance with China’s rigorous censorship regulations.

Instant Cessation: Handling Sensitive Words with Precision

One striking feature of Zhinao is its ability to immediately terminate conversations if a user inputs a “sensitive word.” The company curates a comprehensive list of banned words or phrases, continually updated through the efforts of human moderators and oversight by the public security department. This stringent control mechanism demonstrates how generative AI can adapt to the demands of China’s censorship system.

Also Read:

Invitation-Only Access: Enhancing Content Security

360 Security Technology follows a common restriction policy among Chinese tech giants by providing access to its chatbot through invitation codes. By employing this approach, the company ensures greater control over who utilizes its generative AI product. Content security is touted as a significant advantage of Zhinao, aligning with China’s commitment to maintaining strict oversight over online content.

Also Read: China Sounds the Alarm on Artificial Intelligence Risks

Navigating Local Laws and Ethics

Zhou Hongyi highlighted the importance of adhering to local laws, regulations, ethics, and traditions when developing large language models and generative AI. He emphasized that compliance with these factors is crucial, regardless of the country in which such technologies are developed. Zhou’s statement underscores the significance of aligning AI outputs with local norms and values.

Also Read: U.S. Congress Takes Action: Two New Bills Propose Regulation on Artificial Intelligence

China’s Regulatory Landscape: A Step Towards Controlled Generative AI

In April, the Cyberspace Administration of China (CAC) introduced draft rules governing generative AI. These regulations mandate security assessments for each AI product before public release, prohibiting chatbots from generating content that suggests regime subversion, violence, pornography, or disrupts economic and social order. Although not officially implemented, these rules signify China’s efforts to shape the development & deployment of AI technologies within its borders.

Also Read: EU Takes a Stand with AI Rules

Strategic Evasiveness: Avoiding Sensitive Questions

Chinese ChatGPT-style services, including Baidu’s Ernie Bot, have implemented specific functions to sidestep answering sensitive questions. When asked about China’s democracy, Ernie Bot deftly avoided the topic, stating that it hadn’t yet learned how to answer that question. Such evasive responses demonstrate the careful maneuvering required to comply with China’s strict censorship rules.

Also Read: Ernie Bot vs. ChatGPT: A Comparative Analysis of AI-Language Models

China’s Race to Catch Up: A Positive Outlook

While some experts express concerns that China’s stringent censorship may hinder the development of ChatGPT rivals, Zhou Hongyi remains optimistic. He acknowledged that content moderation was not a unique advantage for 360 Security Technology, as all internet firms in China possess the ability to prevent the generation of banned content. Zhou’s positive outlook suggests China’s determination to catch up with the advancements made by OpenAI’s ChatGPT models.

Embracing the Future: Zhinao and China’s Generative AI Landscape

The unveiling of Zhinao marks a significant milestone in China’s generative AI landscape. The country focuses on ensuring compliance with censorship regulations. Meanwhile, companies like 360 Security Technology strive to develop AI technologies that align with local laws and ethics. The race to catch up with global AI advancements continues, and China remains poised to make remarkable strides in the realm of large language models and generative AI.

Also Read: China’s Billion-Dollar Bet: Baidu’s $145M AI Fund Signals a New Era of AI Self-Reliance

Our Say

360 Security Technology’s Zhinao highlights the intricacies of operating generative AI within a tightly controlled internet environment. As China continues to shape its regulatory framework, the world watches to see how these advancements shape AI’s future. Stay tuned as China pushes the boundaries of generative AI, bridging the gap with global leaders in the field.