The number of ChatGPT users worldwide has surpassed 500 million, fueled by the popularity of the 'Studio Ghibli-style profile picture' created by artificial intelligence (AI). However, concerns are growing about side effects such as copyright infringement and document forgery due to the rapid spread of image generation capabilities.
◇ ChatGPT adds 100 million users in one month
OpenAI announced on its official blog on the 31st of last month (local time) that it has secured $40 billion (about 59 trillion won) in investments from SoftBank Group, and currently more than 500 million people are using ChatGPT. With this investment, OpenAI's valuation is estimated at about $300 billion (approximately 442 trillion won). The company noted, 'Based on the newly secured funds, we will accelerate the expansion of large-scale computing infrastructure and the development of artificial general intelligence (AGI).'
The number of ChatGPT users has surged by 200 million in just three months. It rose from 300 million in December of last year to over 400 million in February this year, and now has surpassed 500 million again. In the past, it took three months to increase by 100 million, but the rate of growth has notably accelerated. In particular, the image generation feature newly integrated into ChatGPT-4o has been analyzed to have sparked user inflow due to its popularity on social media (SNS).
In fact, AI avatar images are flooding SNS due to features that generate 'Studio Ghibli-style profile pictures' and 'Disney and The Simpsons-style characters'. According to Mobile Index, as of the 27th of last month, the daily active user count (DAU) for ChatGPT in South Korea reached 1,252,925, setting a record high.
Recently, OpenAI has significantly relaxed restrictions related to image generation. Previously, it was challenging to create images featuring famous individuals or sensitive topics, but now, most categories including adult content, hate speech, and public figures' images are permitted. Joanne Zhang, head of OpenAI's model behavior, stated, 'While we aim to ensure maximum creative freedom, minimizing real harm is the principle,' adding, 'OpenAI employees should not censor creations individually.'
◇ 'AI must include a marking and watermark'
However, concerns about social issues such as copyright infringement and document forgery are also growing as AI image generation capabilities spread. The process of ChatGPT-4o learning specific animation styles to generate images seems to create a conflict between the protection of creators' rights and the creativity of AI.
Under current copyright law, artistic styles are classified as ideas and do not fall under the protection scope, but there are concerns that if specific characters or scenes are closely replicated, it could lead to infringement. Critics argue that without clear consent or compensation for original creators during AI training, only big tech companies benefit from the profits.
There are also increasing cases where AI image generation technology is being misused for fraud. TechCrunch warned on the 31st (local time) that 'realistic-looking receipts or official documents are being easily generated using ChatGPT-4o.' There have also been confirmed instances of individuals using ChatGPT to create fake restaurant receipts and share them on SNS.
In South Korea, it has recently come to light that a drug offender submitted a fake petition generated by ChatGPT to the court. This suspect faced scrutiny from investigative agencies due to the unnatural writing style and ultimately received a six-month prison sentence in the first trial. The court stated, 'Documents forged by AI technology can severely undermine public trust.'
Kim Myung-joo, head of the Artificial Intelligence Safety Research Institute at the Electronics and Telecommunications Research Institute (ETRI), stated, 'While it's possible to discern that receipts or official documents created by AI are fake upon closer examination, it is not easy for the average person to distinguish them,' adding, 'In order to reduce this confusion, images or documents generated by AI must include a clear mark or watermark indicating they were created by AI.'