SELECT LANGUAGE BELOW

OpenAI seeks to ally election meddling fears in blog post

Artificial intelligence research institute OpenAI published a blog post on Monday seeking to address concerns that its technology could interfere in elections as more than a third of the world prepares to head to the polls this year. .

The use of AI to disrupt election integrity has been a concern since the Microsoft-backed company released two products. ChatGPT, which can convincingly imitate human text, and DALL-E, which can use that technology to create “deepfakes” or realistic images. – Images that appear to be fabricated.

Among those concerned is Sam Altman, CEO of OpenAI, who told Congress in May that through “one-on-one interactive disinformation,” He testified that he was “nervous” about the ability of generative AI to undermine the integrity of elections.

The San Francisco-based company said it is working with the National Association of Secretaries of State, an organization focused on promoting effective democratic processes such as elections in the United States, which holds presidential elections this year.


The use of AI to disrupt election integrity has been a concern since Microsoft-backed companies released ChatGPT and DALL-E. AP

ChatGPT added that it will direct users to CanIVote.org when asked specific election-related questions.

The company is also working to make it clearer that images are generated by AI using DALL-E, and following protocols created by the coalition, it is clear that images are generated by AI. He said he plans to display a “cr” icon to indicate. Content provenance and authenticity.

We are also working on ways to identify content generated by DALL-E even after the images have been modified.


Sam Altman
OpenAI CEO Sam Altman told Congress in May that he was “concerned” that generative AI could undermine election integrity through “one-on-one interactive disinformation.” He testified. AP

In a blog post, OpenAI emphasized that its policies prohibit the use of technology in ways that could be considered potentially fraudulent, such as creating chatbots that impersonate real people or suppressing votes. .

DALL-E is also prohibited from creating images of real people, including political candidates.

The company faces challenges in monitoring what's actually happening on its platform.

When Reuters tried to create an image of Donald Trump and Joe Biden last year, the request was blocked with a message that said it “may not comply with our content policies.”

But Reuters was able to produce images of at least 12 other US politicians, including former Vice President Mike Pence.

Facebook
Twitter
LinkedIn
Reddit
Telegram
WhatsApp

Related News