OpenAI, the company that wants to democratize AI as a chatbot platform for everyone, has decided to put some limits on its service. It has announced a new policy to prevent the misuse of its chatbots for election-related purposes.
The policy, posted on its official blog on the 17th, comes at a time when the world is gearing up for some major political events in 2024. The U.S. presidential election, the Korean general election, and elections in the European Union, Japan, Mexico, Venezuela, Algeria, and Senegal are expected to draw a record number of voters, estimated at 4 billion by The Economist1.
OpenAI said, "We want to make sure our chatbots are used for good, not evil. That's why we have a team of legal, engineering, and policy experts to monitor and address any potential misuse." She added, "We do not allow chatbots that impersonate real people, such as candidates or government officials."
The policy also prohibits any activity that could interfere with the voting process, such as creating apps for political campaigns or lobbying, or spreading false information about voting. Users can report violations to OpenAI.
The move reflects growing concerns about the impact of AI on the integrity and security of elections. Last year, the world witnessed the rise of generative AI, a technology that can create realistic content such as text, images, and audio. This has sparked debate about the ethical and legal implications of AI-generated content, particularly in the context of misinformation and manipulation. A survey by the Organization for Economic Cooperation and Development2 found that "responsible use of technology" and "responding to misinformation and information manipulation" were the top priorities for G7 countries.
The World Economic Forum (WEF), which kicked off on the 15th, also warned that AI could cause electoral chaos as the biggest threat of the year.
Behind these concerns is the rapid popularity of artificial intelligence.
Last year, Generative AI lowered the cost of content production, making it easier for anyone to create and share content. Now, OpenAI is taking it a step further by lowering the barriers to creating and sharing AI chatbots.
The GPT Store, which OpenAI launched on the 10th, is a platform that allows users to create and share custom chatbots without any programming skills. Unlike traditional chatbot development, which requires large amounts of data and programming knowledge, the GPT Store simplifies the process by leveraging ChatGPT, OpenAI's powerful natural language model. The GPT Store has been a huge success, with more than 3 million chatbots shared in less than a week.
The GPT Store offers a variety of chatbots for different purposes, such as DALL-E, writing, research, programming, education, and lifestyle. Some of the featured chatbots include Personalized Trail Recommendations from AllTrails4, Academic Paper Search and Synthesis with Consensus, Coding Skills Improvement with Khan Academy's Code Tutor6, Presentation and Social Post Design with Canva, Book Recommendation with Books, and Math and Science Tutoring with the CK-12 Flexi AI Tutor.
OpenAI also plans to reward chatbot builders based on user engagement with their chatbots. The GPT builder revenue program will launch soon for U.S. builders, according to OpenAI.
In addition to making chatbots easier to build, OpenAI is also making them safer to use. It is developing a program that can reject requests related to real people in image creation, as well as chatbots, and check whether images are AI products. This is to prevent misuse of its image generation model, DALL-E, which can create realistic images from text descriptions.
OpenAI is not the only company concerned about the ethical use of AI. Google also implemented a measure last month to prevent its AI "Bard" from answering election-related questions.
