Rti AI Application Guidelines

Novamber 21, 2024

No. 1130000665

I. Purpose

With the rapid development of Artificial Intelligence (hereafter referred to as AI), its application in media is increasing. However, this may also raise issues related to journalistic ethics, copyright, personal rights, and information security, which must be approached with caution. By referencing international media regulations, these "Rti AI Application Guidelines" (hereafter referred to as “these Guidelines”) have been formulated to provide guidance and a common framework for the use of generative AI in news reporting and program production.

Furthermore, given the rapid evolution of AI, it is essential to continuously observe its development trends, applications, and responses from various media organizations, these Guidelines will be frequently revised accordingly.

II. Basic Principles

(1) When using AI for collaboration, verification, editing, and content publication, it must align with the core values of journalistic ethics, including truthfulness, accuracy, fairness, and objectivity, and give priority to the public interest, as well as comply with all relevant regulations of Rti's “Production and Broadcasting Code”.

(2) A clear distinction between real and artificial content must be maintained. When editing and broadcasting news, priority must be given to content from actual interviews, photographs, and videos taken, as well as audio recordings by real people. It must be remembered that people are the core of news, emphasizing the professionalism, autonomy, creativity, and judgment of each professional, and not replacing their work with AI.

III. Supervision and Review

(1) Clearly define the scope and usage conditions of AI, ensuring that the editorial team can both effectively supervise and implement these guidelines, and reserve the right to discontinue use of AI at any time.

(2) As a support tool, AI is permitted to be used to assist in:

·  Drafting interview questions or program outlines

·  Helping with SEO titles

·  Creating or editing images or charts

·  Assisting with translation

·  Explaining concepts

·  Compiling data

·  Preparing presentations

·  Graphic design

However, AI-generated content such as articles, translated texts, photos, videos, presentations, or design drafts shall not be directly used as final products or broadcast as news or programs without human review, editing, and supplementation.

(3) It is not permitted to use the AI-generated voices or images of Rti's reporters, and anchors, nor shall it be used in news without the prior permission of the supervisor.

(4) In special cases, such as protecting source identities through anonymization, inaccessible locations, extraordinary circumstances, or to help audiences better understand reporting content, AI may be used for voice alteration or generating simulated images only with managerial approval. Additionally, such content must be clearly marked in the report to help audiences identify when it is not a real voice or image.

IV. Public Transparency

(1) When using generative AI as a tool for writing news or as a supplementary tool, appropriate disclosure should be made.

(2) When publishing content that includes AI-generated material, it must be clearly marked in a way easily understood by the audience to facilitate distinction.

(3) Any AI applications that could significantly impact news content production or publication should be clearly disclosed and communicated to the audience.

(4) Maintain records of AI tool usage, including its purpose, scope, and usage conditions, for future verification.

V. Ensuring Content Source Authenticity

(1) Since current AI systems are not entirely transparent, stable and reliable AI tools must be used, and the authenticity and credibility of AI-sourced content must be verified.

(2) Any AI-generated content or queried data may carry risks or be misleading. A cautious and skeptical approach should be maintained, identifying the authenticity of original content sources. One should not completely trust generated information to prevent the spread of misinformation.

(3) When using text or images generated by AI tools, the original content before human modification or reproduction should be retained, and whenever possible, include traceable original data or links for future verification and validation.

VI. Avoiding Bias and Misinformation

(1) Since AI training databases may replicate existing social biases, such as racial or gender discrimination, care must be taken to review the full scope of the database when using generated content to avoid biased or discriminatory material.

(2) News reporting or program information sources should be diverse and should not rely entirely on AI-provided information.

(3) AI governance and use must respect human rights, democratic values, and reflect social and cultural diversity.

VII. Privacy and Confidentiality

(1) Confidential content must be written personally and the use of generative AI is prohibited. Confidential or unpublished content must not be input into generative AI tools.

(2) When entering data into AI systems to generate content, privacy concerns must be carefully addressed and data protection ensured.

(3) Personal data must not be entered into generative AI tools. If necessary, prior consent must be obtained, and information must be encrypted or anonymized before use.

(4) Unpublished official information about the station and contents of contracts with confidentiality clauses must not be input into AI tools.

VIII. Legal Compliance

(1) When using generative AI, privacy rights must be fully respected, and information security, personal data protection, copyright, information usage regulations, and ethical laws must be followed.

(2) Generative AI tools may be complex systems trained on materials used without the explicit permission or knowledge of the original creators. Therefore, it is crucial to take precautions to avoid potential violations of intellectual property rights, trademark rights, design rights, portrait rights, and personal rights.

(3) Generated content – including Images, sounds, videos, and text–  should not be published directly. Human review, along with modifications or reproduction must be conducted before publication to avoid potential copyright concerns.

IX. Accountability

(1) When using AI to collect, process, or distribute information, it is essential to adhere to journalistic ethics and production standards. The editorial team must ensure that all published content is accurate, fair, original, and of high quality.

(2) AI operational processes must be as transparent as possible, and use technically sound and secure AI systems and tools. This approach will help minimize the risk of errors and misuse.

X. Implementation and Revision

These guidelines shall be published and implemented after approval by the Chairman, with the same process for any revisions