© Unsplash/Choong Deng Xiang
This year, many elections will take place around the world. Between the American presidential election and the European elections in the EU, the Axios site indicates that there will be high-stakes elections in 50 countries in 2024. And to protect these elections from misuse of its technologies, OpenAI, the creator of ChatGPT, announces the different measures he will take.
“As we prepare for the 2024 elections in the world's largest democracies, our approach is to continue our platform security work by providing voting information precise, applying measured policies and improving transparency”, explains the organization. “We are deploying a cross-functional effort dedicated to election work, bringing together the expertise of our security systems, threat intelligence, legal, engineering and policy teams to quickly investigate and address potential abuses.”
In its announcement, OpenAI mentions many restrictions that it applies to ChatGPT as well as applications that use its API. At this time, it will not be permitted to use ChatGPT technology to create campaign or lobbying apps. On the other hand, it is prohibited to use this technology to create chatbots that pretend to be real people (such as a candidate). And it is also prohibited to create chatbots based on ChatGPT whose aim would be to discourage voter participation.
For people who get information via ChatGPT, OpenAI indicates that its chatbot increasingly integrates real sources of information. And users will soon have access to real-time information that includes the source of this information, as well as the links. “Transparency around the origin of information and the balance of information sources can help voters better evaluate information and decide for themselves what they can trust,” says the creator of ChatGPT .
And in the United States, OpenAI is collaborating with the National Association of Secretaries of State (NASS), a professional organization, so that ChatGPT can provide correct information about the electoral process. The goal is that if, for example, a user asks where their polling station is, they get correct information.
OpenAI also offers a tool called DALL-E 3, accessible to paying ChatGPT Plus users, which allows you to generate images. It already refuses user requests when they ask to generate images of real people, including those of election candidates. But on top of that, OpenAI wants it to be easy to differentiate between real photos and images generated by DALL-E 3.
And to facilitate this differentiation, it will work with the Coalition for Content Provenance and Authenticity so that the images generated by DALL-E 3 are accompanied by a digital identification. Besides this, OpenAI is developing classification software that detects images generated by DALL-E. This software is currently being tested and the results are said to be promising. Later, this classifier will be made available to more testers, including journalists, online platforms and researchers.
📍 To not miss any news from Presse-citron, follow us on Google News and WhatsApp.< /p>
[ ]
© presse-citron.net If Google focused on artificial intelligence during its I/O conference, it has not…
© Lemon squeezer via 𝕏 If you've owned a few generations of Macs, and whatever…
The new Google Pixel 8a intends to offer many new features initiated by the recent…
Jean Onana and Faris Moumbagna wereé the target on Sunday night à Monday of an…
You want to combine travel and preservation of the planet ? Here are 20 eco-responsible…
The prices defy all competition on this seaside. 3 hours from Paris. Summer approach and…