When AI-generated image of an explosion outside the Pentagon The information, which went viral on social media earlier this week, provided a quick preview of the digital infohazard that AI researchers have warned about for years. Although the image was clearly fabricated, it did not stop prominent accounts such as the Russian state-run RT and @BloombergFeed impersonating Bloomberg News from working with the image. Local police have reportedly received frantic calls from people who believe a new 911-style attack is underway. The ensuing turmoil gave the stock market a brief shock.
The deepfake Pentagon debacle was over in a matter of hours, but it could have been worse. Earlier this year, computer scientist Jeffrey Hinton said:Godfather of AIAs AI-generated images become more and more convincing, he said he fears the public may “be left with no one to know what’s true.”
Startups and established AI companies alike are racing to develop new AI deepfake detection tools to prevent such a reality from happening. Some of these efforts have been ongoing for years, but with OpenAI, generative AI has suddenly exploded into mainstream consciousness. Darui and Chat GPT The sense of urgency to develop an easy way to detect AI deception has led to significant investment.
Companies are racing to find detection solutions across all levels of content. Some companies, such as startup Optic and Intel’s FakeCatch, are focused on exploring AI’s involvement in audio and video, while companies like Fictitious.AI focus on text generated by AI chatbots. We are concentrating our efforts more directly on In some cases, these current detection systems appear to work well, but technical safety experts like former Google Trust and safety chief Arjun Narayan are concerned the tools have not yet caught up. increase.
“The trust and safety industry needs to catch up on how to detect synthetic and non-synthetic media,” Narayan said in an interview with Gizmodo. “As AI progresses, detection technology will probably catch up, but this is an area that needs more investment and further exploration.”
Here are some of the companies leading the deepfake detection race.
