AI Secret: Huang admits relationship with suspected spy; Fuji TV tests AI in its dramas; Accenture and the fabricated quotes

By AI Secret · Mon, 9 Nov 2026.

The newsletter brings together several stories. They focus on Nvidia, the use of generative AI in television and a report with fake citations, among others. The body does not include the URLs of the original articles.

Huang admits to a relationship with a suspected spy. Nvidia CEO Jensen Huang told staff on Sunday that he had cheated on his wife. In doing so, he confirmed reports of a relationship with Alice Huang, a Megaspeed executive whom US intelligence suspects of spying for China. According to the newsletter, Huang wrote: "There is no excuse for what I did".

Nvidia's board is investigating the matter. Three House committee chairs want Huang to testify behind closed doors on November 18. The questions center on chips: Megaspeed bought more than $4.8 billion worth of Nvidia GPUs and is suspected of diverting them to China, which the company denies. Huang called his dealings with Megaspeed "completely normal" and says Nvidia "counted every one" of its GPUs.

According to US intelligence, Beijing wanted her to get Huang to lobby for more sales to China. He had already been openly lobbying for that for years and secured the H200 for China. Trump is backing him: he said Huang "probably has an excuse" and that he would "certainly take him" to Asia. The newsletter notes that the personal matter is new in his 33 years at Nvidia, but the congressional investigation is a rerun of what has been seen before.

Sponsored. Framer is promoting Framer MCP, which connects ChatGPT, Claude, Cursor or Grok to Framer. According to the ad, it lets you ask an agent for a landing page, a blog post or an entire site. The agent builds it on the canvas, populates the CMS and publishes it to the domain, without opening the editor or going through a designer. It works anywhere the AI tool runs, including mobile, and setup takes a minute.

Fuji TV tests AI in its dramas. Fuji Television says all of its dramas will be able to use generative AI by October 2027. It opened an AI lab in September, uses tools such as Google's Nano Banana and offers at least one drama per quarter with an "AI-enhanced" slot. In its current medical series, AI generated a bloody incision and an aerial shot of an explosion that would previously have required a helicopter. Fuji says some steps required up to 80% less work, though not the production as a whole.

The network is setting limits. Shinya Masaki, its head of technology, said: "We are not making everything with AI". Directors still choose, scene by scene, between shooting, using effects or turning to AI.

The newsletter points out that China has already set the precedent. Around 95% of the 128,000 micro-dramas released there in the first quarter were made with AI. Chinese apps account for about 93% of overseas spending on short-drama apps, and Japan is their second-largest market after the US. In the newsletter's view, Fuji's lab serves to save labor, but it also looks like a defense against the Chinese AI dramas already circulating on Japanese phones.

Accenture and the "hand-hallucinated" citations. An Accenture policy report on how AI could help African farmers cites a journal that does not appear to exist, an untraceable group and a blog that now redirects to a Thai gambling site. In all, 15 of its 89 citations were hallucinated. The report was produced together with IE University in Madrid.

Accenture maintains that AI had nothing to do with it. A spokesperson said: "Our review indicates that the errors identified in the report were the result of human error", adding that there is no "evidence" of AI hallucination. The spokesperson did not explain how the company reached that conclusion. The newsletter contrasts this with the company's stance. Julie Sweet, its CEO, said last year that the firm was "exiting" employees it could not reskill for AI. Since February, promotion to leadership roles has depended on "regular adoption" of AI tools.