This year are the presidential elections in the United States and there are national elections in more than 50 countries. They are processes that involve almost half of the population of the planet. An intense electoral calendar that will also occur in full boom of generative artificial intelligence. For this reason, OpenAI, creator of ChatGPT, has gone ahead to announce its strategy to, as it said, “prevent abuses” and provide transparency.
Governments have made clear their concerns about the ability to manipulate and create misleading content through new tools. In fact, disinformation campaigns were recognized as one of the “catastrophic risks” to artificial intelligence in the Bletchley Declaration, signed by 28 countries—including the United States and China—at the summit held last November.
It serves as an example of the commotion caused by the false images of the arrest of Donald Trump – an eventual candidate this year – published in March 2023. Therefore, OpenAI has stated that politicians and their campaigns will not be able to use the company’s artificial intelligence tools. At least, not officially.
“Our focus is to continue our platform security work, raising accurate election information, applying measured policies, and improving transparency,” OpenAI announced in its Blog. The company, a Microsoft partner, explained that will not allow the creation of chatbots that impersonate candidates or government institutions. “We will not allow people to create applications for political campaigns and lobby», he remarked.
OpenAI’s other measures for future elections
The Republican Party in the United States published a video generated by artificial intelligence last year, when President Joe Biden announced that he would run for re-election. “What would happen if the weakest president we’ve ever had was re-elected?”, says the official material published on YouTube. It’s unclear what tools they used, but there are fake images of Biden, wars, and even China invading Taiwan.
«We work to anticipate and prevent relevant abuses, such as deepfakes deceptive, large-scale influence operations or chatbots posing as candidates,” OpenAI highlighted in its statement. He recalled that DALL-E, its image generation tool, has firewalls to reject requests about creating content about real people, such as politicians.
The company will now add digital credentials established by a coalition of artificial intelligence companies. This will encode details about the origin of images created with DALL-E.
OpenAI is also testing a new tool called “provenance classifier”. In this way, he explained, researchers and journalists will be able to detect in the next elections whether an image was generated by artificial intelligence. He hopes to soon make it available to a first group of evaluators.
About ChatGPT, he said that they will continue to integrate their platform with real-time news reports worldwide, including links and clarifications about the source. “Transparency around the origin of information and balance in news sources can help voters better evaluate information and decide for themselves what they can trust,” the statement said.
Against abstentionists
Just as it will try to curb misleading content, OpenAI said it will not allow the development of applications that “dissuade people from participating in democratic processes”. For example, they openly state that voting “doesn’t make sense.”
In the United States, the company created an alliance with the National Association of Secretaries of State, the nation’s oldest nonpartisan organization of public officials. In this sense, ChatGPT will direct users in this country to an authorized website for election information in the United States when they ask questions about how to vote or similar concerns. If this goes well, they could replicate the approach in other countries.
Sam Altman, CEO of OpenAI, had acknowledged last year that the use of intelligence in elections was a “major area of concern.” “I’m nervous about this”Altman said in May of last year, at a hearing before United States senators. During his speech, he advocated for government regulations in this regard.