AWS, as a leader in artificial intelligence (AI) and machine learning, is dedicated to the responsible development and utilization of generative AI. Generative AI, being one of the most groundbreaking innovations of our era, has sparked worldwide fascination. We are fully committed to responsibly harnessing its potential. Our team of responsible AI experts, alongside our engineering and development teams, tirelessly test and evaluate our products and services to address concerns related to accuracy, fairness, intellectual property, appropriate use, toxicity, and privacy. While we may not have all the answers at present, we actively collaborate with others to develop novel approaches and solutions for these emerging challenges. Our aim is to drive AI innovation while implementing necessary safeguards to protect our customers and consumers. At Amazon Web Services (AWS), we understand that generative AI technology and its applications will evolve, presenting new challenges that demand further attention and mitigation. That's why Amazon actively engages with organizations and standard bodies such as NIST, ISO, the Responsible AI Institute, and the Partnership on AI. Recently, Amazon signed voluntary commitments at the White House to support the safe, responsible, and effective development of AI technology.

We are eager to share our knowledge with policymakers, academics, and civil society, recognizing that the unique challenges posed by generative AI require ongoing collaboration. This commitment is in line with our approach to developing our own generative AI services, including the creation of foundation models (FMs) with responsible AI considerations at every stage of the comprehensive development process. From design and development to deployment and operations, we take into account various factors, such as accuracy (ensuring the fidelity of summaries or the factual correctness of biographies), fairness (treating different demographic groups equally in the outputs), intellectual property and copyright concerns, appropriate usage (filtering out requests for illegal activities, medical diagnoses, or legal advice), toxicity (detecting and eliminating hate speech, profanity, and insults), and privacy (safeguarding personal information and customer prompts). All these considerations are integrated into our data training processes, FMs themselves, and the technology used for pre-processing user prompts and post-processing outputs. With regards to FMs, we actively invest in enhancing features and learning from customer feedback as they explore new use cases. For instance, our Amazon Titan FMs have been designed to identify and remove harmful content in customer-provided data, reject inappropriate content in user input, and filter out outputs containing inappropriate content, such as hate speech, profanity, and violence. To aid developers in building applications responsibly, we offer Amazon CodeWhisperer, which includes a reference tracker to display licensing information for code recommendations and provide links to corresponding open-source repositories when necessary. This enables developers to decide whether to utilize the code in their projects and appropriately attribute the source code. Through innovative services like these, we will continue to assist our customers in realizing the benefits of generative AI while collaborating with both public and private sectors to ensure responsible implementation. This collaborative approach will foster trust among customers and the wider public, as we leverage this transformative technology for positive impact. Thank you for your support.