OpenAI Develops High-Accuracy AI Image Detection Tool
OpenAI, the company behind popular chatbot ChatGPT and image generator DALL-E, is working on a new tool to detect images created by artificial intelligence (AI) with a high level of accuracy. According to Mira Murati, OpenAI’s chief technology officer, the tool is currently being tested internally and has a reliability rate of 99%. While no specific timeline for a public release was mentioned, OpenAI is committed to developing ways to identify AI-generated content in images, audio, and text.
The importance of such detection tools is growing as AI can be used to manipulate or fabricate news reports related to global events. Existing tools claiming to detect AI-generated images often lack accuracy. OpenAI had previously released a similar tool in January to detect AI-generated text but discontinued it in July due to its unreliability. The company is now focused on improving the software and expanding its capabilities to identify AI-generated audio and images.
During the Wall Street Journal’s Tech Live conference in Laguna Beach, California, OpenAI CEO Sam Altman and CTO Mira Murati also hinted at the possibility of a new AI model following GPT-4. Although details about a potential successor to GPT-4, possibly named GPT-5, have not been publicly disclosed, OpenAI filed a trademark application for GPT-5 with the US Patent and Trademark Office in July.
Chatbots like ChatGPT, which utilize GPT-4 and its predecessor GPT-3.5, have been known to generate false information, a phenomenon referred to as hallucination. When asked if a GPT-5 model would still produce falsehoods, Murati responded, Let’s see. We’ve made significant progress in addressing the hallucination issue with GPT-4, but there is still more work to be done.
In addition to image detection, OpenAI’s focus on AI-generated content aligns with Adobe Inc.’s Firefly image generator. Adobe promises its tool will not create content that infringes on intellectual property rights. This concerted effort aims to combat the spread of misleading information and protect original content.
Altman also addressed the possibility of OpenAI designing and manufacturing its own computer chips for training and operating its AI models instead of relying on existing providers like Nvidia Corp. Although it is currently not the default path for OpenAI, Altman did not rule out the potential for in-house chip development in the future.
As OpenAI continues to pioneer advancements in AI technology, their image detection tool represents a significant step forward in combating the proliferation of AI-generated content. With further improvements and developments on the horizon, OpenAI aims to ensure the accuracy and reliability of AI-generated images, audio, and text.