Deepfakes are projected to become so sophisticated by 2026 that they will be virtually indistinguishable from authentic recordings for the average person, according to a computer scientist researching synthetic media. The rapid advancement in artificial intelligence (AI) has led to AI-generated faces, voices, and full-body performances mimicking real people with increasing accuracy, exceeding expert expectations in recent years.

These deepfakes are not only improving in quality but also increasing in volume. Cybersecurity firm DeepStrike estimates that the number of online deepfakes grew from approximately 500,000 in 2023 to around 8 million in 2025, representing an annual growth rate of nearly 900%. This surge poses a significant challenge to discerning reality from fabrication, especially in low-resolution video calls and social media content.

The researcher, who specializes in deepfakes, anticipates that the situation will worsen in 2026 as deepfakes evolve into synthetic performers capable of real-time interaction. This development could have far-reaching implications across various sectors, including entertainment, politics, and cybersecurity.

Voice cloning, a key component of deepfake technology, has reportedly crossed the "indistinguishable threshold," making it easier to create convincing audio impersonations. This advancement raises concerns about potential misuse, such as fraudulent schemes and the spread of misinformation.

The rise of deepfakes necessitates increased awareness and the development of tools to detect synthetic media. Experts recommend verifying information from multiple sources and being cautious of content that seems too good or too outrageous to be true. As deepfake technology continues to advance, the ability to identify and mitigate its potential harms will become increasingly crucial.