A breakthrough in AI image generation has been announced by the InstantX team in Beijing. The new method, called InstantID, is capable of quickly recognizing individuals and generating new images based on a single reference image. Reuven Cohen, an enterprise AI consultant, hailed InstantID as the ‘new state-of-the-art,’ but expressed concerns about its potential to fuel a surge in deepfake audio, images, and video tools, particularly with the upcoming 2024 election.
Cohen highlighted the ease of creating highly realistic deepfakes using InstantID, without the need for extensive training or fine-tuning, and with minimal GPU and CPU resources. InstantID’s capabilities surpass those of LoRA, a model known for its AI image generation, which has previously led to the proliferation of deepfakes across various platforms.
InstantID’s arrival has prompted discussions within the AI community, with Cohen declaring ‘So long, LoRA’ and referring to InstantID as ‘deep fakes on steroids.’ The potential impact of this advancement in AI image generation is significant, raising concerns about the authenticity and trustworthiness of digital content in the future.
