Korea probes AI agents in bank hacks as president cites 'signs'
South Korea opened a formal investigation into whether AI agents drove a wave of bank breaches — and it isn't the only AI story moving money today.
President Lee Jae Myung said "signs" point at AI models, and the probe is now at the highest level a national banking sector has seen. Speaking at a cabinet meeting, Lee said that "in some hacking incidents, signs have emerged of AI being used, causing considerable public concern and anxiety," and police have since opened a full-scale investigation. The breaches landed on Shinhan Bank, KB Kookmin, Hana and Woori in a campaign that began September 28, and Korean media citing the country's financial security institute report seven institutions hit, more than 65,000 customer records exposed, and an open-source autonomous penetration-testing tool called ARTEX traced via IP and log analysis — with intruders dwelling 30 to 43 hours before detection. Why it matters: this is the first case of a regulator-confirmed AI-agent attack tool working against a real national banking sector, escalated to head-of-state level. The uncomfortable pattern is that agentic attacks are no longer demos — the tooling is off-the-shelf, and the detection gap measured in tens of hours is the actual story.
ChatGPT isn't just imitating New Yorker cartoonists — it's signing their names to the fake cartoons. An investigation published by Nieman Lab documents at least 15 cartoonists whose real signatures ChatGPT reproduced on generated cartoons, one of which drew roughly 25,000 likes in a single post; OpenAI said it appreciates "community flagging bugs and unintended behavior," but the piece found signatures still appearing after notification. The New Yorker's parent, Condé Nast, says its 2024 license with OpenAI never permitted cartoon training, and the cartoonists' contracts grant no AI rights — while cartoonist Brendan Loper's protest ("my name is my name. It felt very much like a violation of my personhood") points at the right-of-publicity claim style itself can't support. The distinction between copying a style and forging a signature just became the sharpest line in the AI copyright debate.
Kling has picked banks for a billion-dollar Hong Kong IPO — China's AI video leader is going public. Bloomberg reports Kuaishou's AI video unit hired CICC, Goldman Sachs and UBS to raise at least $1 billion in a listing targeted for as soon as next year, after a July round from Alibaba, Tencent and Baidu valued Kling at roughly $15 billion pre-money; shares of Kuaishou rose about 3% on the news. Chinese media put the post-money valuation near 120.6 billion yuan (about $18 billion), with second-quarter revenue above 850 million yuan, up 200% year over year. Kling follows Moonshot through the same door — we tracked Moonshot AI files for Hong Kong IPO at $50B — and the read is that China's AI labs are converting valuations into public listings before the window narrows.
What to watch: whether Korean authorities name the specific AI tools involved — the hedge around "signs" suggests they haven't finished attribution.
Should labs be on the hook when off-the-shelf open-source agents are used to break into banks? Tell us in the comments.
