OpenAI Launches New Text-to-Image Generator with Safety Measures

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

Overview

OpenAI is expanding access to its latest text-to-image generator, DALL-E 3. Customers of ChatGPT Plus and Enterprise will now be able to use the new model within the ChatGPT app. OpenAI claims to have implemented safety measures to ensure a more responsible and secure user experience.

Improvements Over Previous Model

DALL-E 3, which was announced last month, builds upon the capabilities of DALL-E 2. This new model allows users to create longer and more detailed prompts using ChatGPT, resulting in more visually descriptive images. Microsoft’s Bing platform was the first to offer public access to DALL-E 3, even before ChatGPT.

Controversy and Safety Concerns

Previous text-to-image generators have faced controversies and safety issues. Users have generated inappropriate and misleading images, including copyrighted materials, nonconsensual nudes, and altered representations of public figures. Despite attempts to address these issues, some users have found workarounds.

Enhanced Safety Measures

OpenAI claims to have taken extensive steps to mitigate potential harms with DALL-E 3. The company provides a website showcasing its research on the model and has implemented guardrails to limit the generation of certain types of content. These guardrails aim to prevent the model from creating images resembling the work of living artists, images of public figures, and to improve diversity in the generated images. OpenAI also developed an internal tool called the “provenance classifier” that can accurately detect if an image was generated by DALL-E 3 with 99 percent accuracy.