San Francisco, Dec 23 – The AI Foundation Model Transparency Act, introduced by two US lawmakers, could potentially require artificial intelligence (AI) companies to disclose copyrighted training data. Representatives Anna Eshoo (D-CA) and Don Beyer (D-VA) filed the Bill, which aims to establish rules for reporting transparency related to the usage of copyright data for training AI models.
If passed, the Federal Trade Commission (FTC) would collaborate with the National Institute of Standards and Technology (NIST) to develop guidelines regarding the public disclosure of training data used in AI foundation models. The ultimate goal is to ensure transparency and accountability within the AI industry.
The proposed legislation would necessitate companies building foundation models to divulge the sources of their training data and outline how this data is preserved during the inference process. Additionally, they would be required to describe any limitations or risks associated with their AI models.
In an effort to prevent the dissemination of inaccurate or potentially harmful information, AI developers must report their endeavors to red team the model. This process involves testing the model extensively to identify and rectify any inaccuracies specifically in sensitive domains such as medical and health-related queries, biological synthesis, cybersecurity, elections, policing, financial loan decisions, education, employment decisions, public services, and vulnerable populations like children.
The rationale behind the legislation stems from the surge in copyright infringement concerns and public apprehensions surrounding the use of foundation models. The increased accessibility of artificial intelligence has led to instances where inaccurate, imprecise, or biased information was presented to the public.
If implemented, this Bill would play a crucial role in safeguarding against the potential pitfalls of utilizing AI models without clear visibility into the training data and algorithms employed. It aims to strike a balance between technological advancements and public interest, ensuring that AI companies are held accountable and minimizing the risk of misinformation or unethical practices.
The proposed regulations also highlight the significance of establishing standards and guidelines to govern the growing AI industry. As AI continues to infiltrate various sectors, it is crucial to institute proper frameworks and systems that foster responsible AI development and deployment.
While the Bill awaits assignment to a committee, it signifies a significant step towards addressing the challenges associated with AI transparency. As lawmakers recognize the importance of public access and understanding in the realm of artificial intelligence, this initiative seeks to create an environment where AI models operate with integrity, promoting trust and fostering fair practices for the benefit of society as a whole.
