OpenAI is taking measures to ensure the responsible use of its powerful language and image models as the 2024 elections approach. With over 180 million users, the AI lab’s products, including ChatGPT and DALL-E, have the potential to be misused for influencing election campaigns. To address this, the company has outlined its strategy for the upcoming elections in a blog post on January 15th.
Preventing Abuse of OpenAI’s Systems
OpenAI is proactively working to strengthen its AI systems against exploitation by malicious actors. This includes thorough testing, gathering user feedback during development, and building protective measures into the foundation of its models. For example, DALL-E, the image generator, has strict policies that reject any image generation requests involving real people, particularly political candidates.
The company has stated that it is focused on anticipating and preventing relevant abuse, such as misleading ‘deepfakes’, scaled influence operations, or chatbots impersonating candidates. Additionally, usage rules prohibit ChatGPT applications for propaganda, voter suppression tactics, or political impersonation bots.
Snapshot of how we’re preparing for 2024’s worldwide elections:
• Working to prevent abuse, including misleading deepfakes
• Providing transparency on AI-generated content
• Improving access to authoritative voting informationhttps://t.co/qsysYy5l0L— OpenAI (@OpenAI) January 15, 2024
Humans Brought into the Fold
In a surprising move, OpenAI plans to involve humans in the fact-checking process by utilizing new transparency features to trace AI creations back to their origins. Digital watermarking and fingerprinting will verify DALL-E images, and news links and citations will be more prominently displayed within ChatGPT search responses. Additionally, the company aims to collaborate with nonpartisan voting agencies like the National Association of Secretaries of State (NASS) in the US and provide official registration and polling details from CanIVote.org to counter misinformation.
While these measures are widely supported, OpenAI acknowledges that bad actors will continually attempt to misuse these tools for electoral purposes. The company is positioning itself to adapt to the challenges it faces during election cycles, and it anticipates that collaboration with Big Tech and governments may provide the most sustainable approach to combat AI fakes and propaganda.
Featured Image: Unsplash
