Close
1. Purpose
Baidu, as a leading global AI enterprise, has always regarded AI ethics as the cornerstone of its development. Adhering to the vision of "making the complex world simpler with technology," Baidu is committed to building an AI ecosystem that is safe, controllable, and promotes the equal growth of humanity. We are dedicated to providing guidance and consensus for all company personnel and business, ensuring they fulfill their social responsibilities and clarify the direction for technological innovation, ensuring that AI technology brings more freedom and possibilities to humanity.
Baidu's AI ethics principles and initiatives complement the company's code of conduct, human rights policies, security policies, and information security policies, forming a comprehensive system that ensures the company's AI research and development and practices meet the requirements and expectations of all stakeholders regarding the company's AI ethics performance.
2. Principles
In accordance with the international AI ethics framework, Baidu has formulated AI ethical principles and developed Baidu's Six Initiatives on Science and Technology Ethics.
2.1 Four Principles of AI Ethics
Baidu has formulated its AI ethics principles emphasizing four key values of safety, equality, empowerment, and freedom. These principles specifically include:
2.2 Baidu's Six Initiatives on Science and Technology Ethics
Baidu has referred to mainstream international AI ethics frameworks to establish its initiatives based on the widespread consensus on safety control, human-centricity, inclusiveness, advancement promotion, and privacy protection. Building on Baidu's four AI ethics principles, the company has formally proposed the Six Major Initiatives for Baidu's Technology Ethics, which include:
First, manage ethical risks and promote innovative development.
Second, human and machine values must be aligned to ensure that technology serves humanity.
Third, protect individual privacy and ensure data security.
Fourth, expand capability openness and commit to the widespread benefit of technology.
Fifth, strengthen awareness cultivation and respect for ethical morals.
Sixth, communication and cooperation should be enhanced to achieve diverse governance.
3. Implementation
3.1 Establish the Technology Ethics Committee
To manage ethical risks and promote innovative development, in October 2023, Baidu established a Technology Ethics Committee, with Li Zhenyu, Senior Vice President and CEO Assistant of Baidu Group, serving as the chairman of the committee, and Chen Shangyi, Chairman of the Baidu Technology Committee, serving as vice chairman. At the same time, the committee gathers the strength of experts from various fields, introduces authoritative external perspectives and supervision, and earnestly fulfills the main responsibility of technology ethics management.
3.2 Conduct AI Ethics Training
Baidu regularly conducts AI ethics-related training sessions, inviting experts and scholars to deeply discuss courses on the legality and compliance of AI large models, fairness and non-bias during operation, and AI accountability mechanisms. The AI ethics training is aimed at employees in various positions within Baidu, with the goal of deepening the understanding of technology ethics from multiple perspectives, providing a more solid ethical foundation for the company's technological development and strategic decision-making, and leading the future direction of technological development. At the same time, Baidu provides relevant training to suppliers according to the Baidu Sustainable Development Principles for Suppliers to enhance their awareness and literacy in AI ethics.
3.3 Participate in the Development of AI Ethics Standards
Baidu actively participates in the development of standards related to artificial intelligence technology and ethical governance. It continuously shares advanced technologies and practices with the industry to form industry consensus and is committed to building a healthy and trustworthy ecosystem for the development of the artificial intelligence industry.
3.4 Develop Explainable AI
Baidu continuously strengthens safety assessment and control capabilities during the R&D process to ensure that AI is safe and controllable. It also actively participates in the national ethical governance process, making it a basic requirement that technological development "conforms to public morality and ethics." This approach helps to clarify the boundaries for technological innovation, effectively prevent major technological ethical risks, and further create a responsible and sustainable AI.
3.5 Establish a Safety Assurance Mechanism
Baidu has formulated internal regulations such as the Baidu Data Classification Specification, the Detailed Rules on Handling of Security Events of Baidu, the Baidu Data Security Emergency Response Process, the Data Security Emergency Plan, and the Baidu Information and Product Security Punishment Measures. These regulations standardize the definition and grading of security incidents, the safety management responsibilities at various levels, and the penalty mechanisms for security incidents at different levels, ensuring that security incidents are seriously dealt with according to their severity. The specific safety emergency response process includes detection and response, analysis and reporting, recovery and review.
Baidu conducts internal and external security audits annually to ensure the effectiveness and reliability of its safety management system. The audit content includes the company's data security, personal information protection, overall governance structure, and data classification and grading.
4. Stakeholder Engagement
Baidu values the expectations, support, and participation of stakeholders, including shareholders, employees, users, customers, suppliers, government departments, social organizations, media, and communities, in the process of implementing this AI ethics policy. Therefore, we welcome all internal and external stakeholders to maintain communication and cooperation with us regarding our AI ethics policy and AI ethics practices.
