Microsoft and OpenAI Start $2M Fund to Fight Election Deepfakes
Microsoft and OpenAI have announced a $2 million fund to combat the growing risks of using AI and deepfakes to “scam voters and weaken democracy.”
There are worries about the impact that AI will have on voters, especially those in “vulnerable communities,” who are more likely to believe what they see. This year, a record 2 billion people will go to the polls in elections in about 50 countries.
With the rise of generative AI, like the wildly popular chatbot ChatGPT, there are now a lot of “deepfakes” made by AI that are meant to spread false information. The ease of access to these new tools, which enable anyone to create phony videos, photos, or audio featuring well-known political figures, is detrimental.
India’s Election Commission told political parties on Monday not to use deepfakes and other false information in their online campaigns leading up to the elections.
Because of this, all of the big tech companies, like Microsoft and OpenAI, have signed voluntary pledges to protect against these kinds of risks. Additionally, they aim to collaborate in developing a shared framework to address deepfakes designed to deceive voters.
In other places, big AI companies have started to deal with these risks by making their software less flexible. For example, Google has said that its Gemini AI chatbot won’t be able to answer questions about elections. Meta, the company that owns Facebook, is also limiting how its AI chatbot can answer questions about elections.
OpenAI released a new deepfake detector earlier today to help researchers looking into disinformation find fake content made by its own DALL-E image generator. The company also joined the steering committee for the Coalition for Content Provenance and Authenticity (C2PA), an industry group that already has Adobe, Microsoft, Google, and Intel as members.
The new “societal resilience fund” is part of this larger push for “responsible” AI. According to a blog post the companies put out today, they want to “further AI education and literacy among voters and vulnerable communities.” The Coalition for Content Provenance and Authenticity (C2PA), the Older Adults Technology Services (OATS) group, the International Institute for Democracy and Electoral Assistance (International IDEA) group, and the Partnership on AI (PAI) group are among the organizations that could receive grants for this.
Microsoft, a software giant, claims that these grants aim to enhance everyone’s understanding of AI and its capabilities. For example, OATS will reportedly use its grant to fund training programs in the United States for people aged 50 and up that cover the “basics of AI.”
Microsoft and OpenAI are committed to addressing the problems and needs in AI education and literacy. The Societal Resilience Fund’s launch is just one step in that direction. “Microsoft and OpenAI will continue to work on this project, as well as with groups and projects that share our goals.”