In January of this year, OpenAI unveiled revamped policies aimed at preventing its tools from being used to spread disinformation ahead of the 2024 election, in an attempt to prevent people from building chatbots "for political campaigns and lobbying", especially when it comes to targeting Latino communities in the U.S.

However, a study titled "De(generating) democracy? A look at the manipulation of AI tools to target Latino communities online", conducted by the Digital Democracy Institute of the Americas and published in April, reveals that even with policies established by OpenAI, users with minimal knowledge can circumvent restrictions to target specific political content to Latino voters. The experiments carried out have shown that both "ChatGPT" and the image generator "DALL-E", can be used for this purpose, despite policies aimed at preventing such misuse.

Additionally, the study highlights that OpenAI's content policies, while advanced compared to other AI companies, still have loopholes that allow for the manipulation and targeting of political content to minority and marginalized communities in the US.

In this sense, the research realizes the need for tech companies, such as OpenAI, to do more to ensure equitable enforcement of their policies in all communities that use their tools. This is especially relevant considering the rapid growth and popularity of OpenAI's ChatGPT, which has a large number of users and can potentially be exploited for a variety of political purposes.

The OpenAI spokeswoman's statement highlights a disagreement between the report's findings and the company's understanding of its own policies and tools. While OpenAI claims to allow the use of its products for political advocacy, the study indicates that the company's tools still offer guidance that can be exploited to create targeted political content.

In this context that precedes electoral disputes, the complexity and challenges faced in regulating and enforcing policies to prevent the misuse of AI tools in political campaigns and the spread of disinformation are visible, especially when it comes to targeting specific communities, such as Latino voters in the US.