Elon Musk’s AI firm xAI is on the heart of an argument over the unfold of deepfake pictures. AFP-Yonhap
As Elon Musk’s synthetic intelligence firm xAI faces backlash over deepfake pictures, Korea’s Framework Act on the Improvement of Synthetic Intelligence is about to take impact subsequent week. All eyes are on whether or not this laws — the world’s first complete AI regulation — can successfully fulfill its position.
Nevertheless, skepticism is rising concerning the regulation’s effectiveness, with many predicting it is going to face sensible hurdles in blocking xAI’s deepfake companies or limiting entry.
Trade officers mentioned Tuesday that xAI’s AI mannequin, Grok, continues to supply options on Musk’s social media platform X that flip on a regular basis pictures into sexually specific deepfake pictures.
Whereas Malaysia and Indonesia have restricted entry to the platform and different nations have launched authorized investigations — prompting xAI to restrict the characteristic to paid subscribers — issues stay.
The act, which takes impact Jan. 22, was enacted to strengthen AI operators’ tasks and create a basis geared toward fostering belief in an AI-driven society. Stopping deepfake crimes has been cited as a key purpose.
The regulation applies to abroad operators and requires them to designate home brokers to satisfy authorized obligations.
Nevertheless, the transparency obligation to label deepfakes will not be particularly included within the home agent’s position. This ambiguity may delay prevention or response when speaking with overseas operators.
As a substitute, the act itself mandates that AI-generated content material have to be labeled. It requires seen watermarks on deepfake content material which are laborious to tell apart from actuality. Violations set off a correction order, and failure to conform can lead to a superb of as much as 30 million received ($20,300).
Underneath these guidelines, Grok’s deepfake pictures should carry watermarks. However for the reason that act features a minimal one-year grace interval, rapid enforcement is inconceivable.
Analysts say that even after the grace interval ends, instantly blocking or limiting overseas AI companies will stay tough attributable to potential commerce friction.
“Underneath present legal guidelines, it’s laborious to do greater than impose fines if abroad firms like xAI don’t cooperate voluntarily,” mentioned Jung Chang-woo, a lawyer at Lee & Ko.
Consultants advise that till the brand new AI act takes full impact, deepfake circumstances ought to be dealt with underneath current legal guidelines such because the Data and Communications Community Act or the Private Data Safety Act.
“The binding pressure of the act alone is weak,” mentioned Yeo Hyun-dong, a lawyer at Yoon & Yang LLC. “Laws have to be supplemented by sanctioning particular violations at the side of current legal guidelines.”
The federal government maintains its stance on making use of minimal regulation whereas monitoring the scenario for a sure interval. A Ministry of Science and ICT official mentioned, “As AI expertise remains to be creating by trial and error, we are going to watch the scenario for now to permit for self-correction.”
This text from the Hankook Ilbo, the sister publication of The Korea Instances, is translated by a generative AI system and edited by The Korea Instances.
