OpenAI has outlined its plan to combat misinformation in the lead-up to elections worldwide, emphasizing transparency regarding the sources of information. The company is taking measures to prevent abuse, enhance transparency for AI-generated content, and facilitate access to accurate voting information.
OpenAI's cross-functional team, comprising experts from safety systems, threat intelligence, legal, engineering, and policy, is dedicated to addressing potential abuses related to elections. They aim to prevent misleading "deepfakes," scaled influence operations, and chatbots impersonating candidates.
To ensure transparency in AI-generated content, OpenAI is working on provenance efforts. In early 2024, it plans to implement the Coalition for Content Provenance and Authenticity’s digital credentials for images generated by DALL·E 3. This approach encodes details about the content’s provenance using cryptography. Additionally, OpenAI is experimenting with a provenance classifier, a tool designed to detect images generated by DALL·E.
As the US approaches its upcoming presidential election, OpenAI is collaborating with the National Association of Secretaries of State (NASS). For example, ChatGPT will direct users to CanIVote.org, an authoritative website on US voting information, when asked specific procedural election-related questions.
The company is committed to red-teaming new systems, seeking feedback from users and external partners, and implementing safety mitigations before releasing any new technology to minimize the potential for harm. These initiatives collectively aim to create a more secure and transparent environment during election periods.
(With Agency Inputs)
Read also| OpenAI Hits Financial Milestone in 2023 with a 700% Revenue Surge: Report
