Advertisement
Advertisement
Artificial intelligence
Get more with myNEWS
A personalised news feed of stories that matter to you
Learn more
The sudden share price swing that hit iFlytek this week underscores the potential cybersecurity risks posed by generative artificial intelligence systems like ChatGPT and similar chatbot services. Photo: Shutterstock

Chinese artificial intelligence firm iFlytek blames chatbot-generated article for sudden share price swing on Shenzhen bourse

  • A chatbot-generated article about how iFlytek gathers user data for AI research caused an unexpected drop in its Shenzhen-listed shares on Wednesday
  • The firm has threatened legal action on those behind the article, but did not make any comment about which Chinese chatbot was used to create it
Chinese artificial intelligence (AI) company iFlytek, one of the first major tech firms in the country to launch a ChatGPT alternative, has blamed an online article generated by an unnamed local chatbot for the sudden decline of its Shenzhen-listed shares on Wednesday, an incident that could become a portent of potential risks from using this technology.

Shares of iFlyTek, which have risen more than 70 per cent since the start of this year, dropped more than 9 per cent in mid-day trading on Wednesday before closing down 4.26 per cent to 56.57 yuan (US$8.02). On Friday, the company’s shares closed up 1.83 per cent to 56.63 yuan.

In two separate inquiries made via iFlytek’s investor relations website on Wednesday, retail investors suggested that the unexpected price swing was brought about by a widely-circulated online article about how the company illegally collects user information for its AI research. They said the article was later found to be AI-generated.

The company said it agreed with the investors’ finding, but did not make any comment about which Chinese chatbot was used to generate the offending article.

Liu Cong, an iFlytek vice-president and head of its research operations, is seen on stage and on the large screen at the Chinese artificial intelligence firm’s launch of its SparkDesk chatbot in Hefei, capital of eastern Anhui province, on May 6, 2023. Photo: Handout

“It’s illegal to use large language models [LLMs] to generate falsehoods, a behaviour that will be punished by law,” an iFlytek representative said. “We will take legal action against any entity or individual who fabricates false information to stigmatise our company.”

The company did not immediately respond to a request for comment on Friday.

An LLM is a deep-learning algorithm that can recognise, summarise, translate, predict and generate text and other content based on knowledge gained from massive data sets, according to US tech firm Nvidia, which supplies chips that power AI systems.
The sudden share price swing that hit iFlytek underscores the potential cybersecurity risks posed by generative AI systems like ChatGPT and similar chatbot services.
Visitors at iFlytek’s SparkDesk launch, held in the eastern city of Hefei, check out features of the Chinese artificial intelligence firm’s chatbot on May 6, 2023. Photo: Handout
In February, Beijing police issued a public warning about rumours generated by ChatGPT. It said research institutes have found that the AI chatbot, when asked conspiracy-related and misleading questions, quickly generated compelling information without citing sources.
In the same month, a resident of Hangzhou, capital of eastern Zhejiang province, used that chatbot to generate a post resembling an announcement from the city that it would lift number plate-based driving restrictions, a measure many Chinese cities have implemented to reduce traffic congestion.

The iFlytek representative on Wednesday told investors that the company’s user privacy protections include a dedicated committee that oversees all privacy matters, which ensures the firm complies with national standards.

China’s internet watchdog warns of risks from generative AI, other advanced tech

Based in Hefei, capital of eastern Anhui province, iFlyTek launched its own AI chatbot, called SparkDesk, earlier this month. It claimed that SparkDesk will exceed ChatGPT’s Chinese data capacity and catch up on English-language resources by the end of October.

That declaration made iFlytek, which has been on the US government’s trade blacklist since October 2019, the first Chinese tech company to provide a clearly articulated timeline to challenge ChatGPT.
Baidu, meanwhile, announced last week that it has started to offer Ernie Bot, its own ChatGPT alternative, to a variety of clients in industries such as financial services, software and education.
Post