China Takes on Student Cheating by Shutting Off AI Nationwide During Exams
How do you make sure that kids aren't using artificial intelligence tools to cheat on their final exams? Easy, just flip the off switch. According to Bloomberg, Chinese AI companies have shut down their AI chatbots, rendering them unavailable while students across the country take their annual college entrance exams.
Popular chatbots including Alibaba's Qwen and ByteDance's Doubao have blocked photo recognition features from being used to identify and answer questions related to the test, while Tencent's Yuanbao and Moonshot's Kimi have cut access to photo recognition tools entirely during dedicated exam hours. The Guardian reported the same is true of DeepSeek.
When students ask the chatbots why they can no longer use the features, they are told, 'To ensure the fairness of the college entrance examinations, this function cannot be used during the test period,' per Bloomberg. If they try to upload an image or ask for help solving an exam question, the chatbots inform them it is 'not in compliance with rules.'
While the shutdown of AI features seems to be basically universal across all companies offering such services, it doesn't appear that any of the companies made public announcements of these restrictions. Instead, according to The Guardian, word of the lack of availability has been driven primarily by students on social media—many of whom are freaking out about their inability to use the tools—and other users annoyed by being cut off from the features. 'College entrance exam candidates, you are all shit,' one Weibo post spotted by The Guardian read. 'I can't use DeepSeek to upload pictures, I have to download ChatGPT again, I hope you all go to community college.'
The exam period for these students in China is no joke. The exam, known as 'gaokao' is a three-day affair that high school students are put through in hopes of getting a spot at one of the country's universities, which have limited spots available. An estimated 13.3 million students will take the exams this year. The exam lasts about nine hours, split across three days, and students are not allowed to use their phones or laptops during the testing period. (That is probably enough to prevent kids from accessing AI tools, but China isn't taking any chances there.)
While the kids can't use AI during the exam, it seems the administrators of the test can. China Daily reported that some testing sites are utilizing AI surveillance systems that are designed to flag 'irregular behavior' like students whispering to one another or making repeated glances that might not be spotted by the human proctor.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Washington Post
27 minutes ago
- Washington Post
Office of the CTO Announces Sam Han as Chief AI Officer and Creation of WP Incubator
In line with our plan to create innovative products to better serve and reach new customers, I am pleased to announce the promotion of Sam Han to Chief AI Officer and separately, the creation of WP Incubator. As our industry undergoes significant changes, we are moving with agility to meet audiences where they are and anticipate the needs of consumer preferences. Sam's promotion and the launch of the WP Incubator are important elements of our overall strategy as we continue to lead the industry with AI innovation. Sam's new role is a first for us and underscores our dedication to leading the industry in this space. An experienced AI practitioner with a Ph.D. in Computer Science, Sam has been with the company for nearly eight years and has an in-depth understanding of AI and Machine Learning. Sam will work with teams across the company to continue to lead the development of The Post's AI suite of products as we continue our transformation. WP Incubator, modeled after successful incubators in Silicon Valley, has been designed to develop cutting-edge products and potential new business lines. It will identify and cultivate new ideas and accelerate their journey from concepts to market-ready products. To support WP Incubator, we have created new roles that align with our goals and help us stay ahead in this rapidly evolving industry. Norman Winarsky, a respected figure in the innovation space, will join as an advisor, bringing invaluable insights and expertise. Lippe Oosterhof, who is already a Strategic Advisor to The Post and media veteran with 20 years' experience, will join as the first Entrepreneur-in-Residence (EIR), further strengthening the incubator's talent pool. While we move with speed in service to our mission, our commitment to ensuring fact-based and informative information does not waver. We will continue to ensure we live up to the standards and ethics needed to preserve and produce high-quality journalism and riveting storytelling. I am proud of the work that we have achieved together in the last two years, and thank you all for your great work and dedication to meet this moment. I look forward to building on our momentum with these two strategic developments. The WP Incubator and Sam as Chief AI Officer will be part of the Office of the CTO. Please join me in congratulating Sam and welcoming Norman and Lippe.
Yahoo
28 minutes ago
- Yahoo
Arkansas Supreme Court releases proposed rule for artificial intelligence
The Arkansas Supreme Court building in Little Rock. (John Sykes/Arkansas Advocate) The use of artificial intelligence in legal documents could violate Arkansas law or court rules, according to a proposed administrative order issued by the state Supreme Court last week. Specifically, the proposed order addresses the use of confidential court data with generative artificial intelligence. AI models retain data inputted by users of AI products, such as ChatGPT, in order to continue training the large language models that exploded into public use only a few years ago, the order notes. 'Anyone who either intentionally or inadvertantly [sic] discloses confidential or sealed information related to a client or case [to a generative AI model] may be violating established rules,' the proposed order reads, specifically citing Arkansas Supreme Court Administrative Order Number 19, the Arkansas Rules of Professional Conduct and the Arkansas Code of Judicial Conduct. Additionally, the proposed order prohibits anyone with internal access to the state's court system, CourtConnect, from 'intentionally exposing our state courts' internal data to a GAI.' The proposed order provides an exemption to this prohibition if approval is granted by the Supreme Court's Automation Committee to engage in 'a research and analysis project related to the use of generative AI tools and general AI for the benefit of our courts.' The proposed order does not appear to address questions of broader use of AI by attorneys within the state court system. Judges in courtrooms across the country in recent months have expressed frustration with attorneys who have filed briefs and other documents bearing citations to nonexistent or irrelevant cases as a result of so-called 'AI hallucinations,' leading to sanctions in some cases. As reported by the Alabama Reflector, for example, lawyers who were being paid millions by the Alabama Department of Corrections to defend it against lawsuits filed by prisoners in the state system were called out by an inmate's attorneys for making up legal citations 'out of whole cloth' in a lawsuit where their client alleged being stabbed repeatedly while in restraints. The federal judge presiding over the case said that the incident showed that sanctions levied by other courts had proven 'insufficient' to deter lawyers from filing documents with improper or made up citations created by AI. 'That causes me to consider a fuller range of sanctions,' Judge Anna M. Manasco said. The Arkansas Supreme Court Committee on Automation created a subcommittee to 'study the use of AI in the courts.' The introduction to the proposed order notes that as the committee continues its work, it will make recommendations. The comment period for the proposed administrative order ends on Aug. 1. SUBSCRIBE: GET THE MORNING HEADLINES DELIVERED TO YOUR INBOX
Fast Company
39 minutes ago
- Fast Company
Are we going to have a recession? CEOs don't think so—in fact, many say the economy will grow
More American CEOs are optimistic about the nation's economy ahead of July's major trade deadline than they were in the last few months, following the beginning of President Donald Trump's tariff wars, according to this month's CEO Confidence Index survey from the Chief Executive Group. On Monday, the US and China restarted trade talks in London, following talks between Trump and China's Xi Jinping last week. The U.S. is trying to speed up negotiations before Trump's 'reciprocal' tariffs go back into effect on July 9 after a 90-day pause. CEOs and business leaders in the United States are also less likely to say the country is headed toward a recession, according to the survey data. While some 62% of CEOs predicted a recession within 6 months back in April, now less than half of that, or 30%, forecast either a mild or severe recession over the next six months. The latest CEO Confidence Index survey polled 277 U.S. CEOs just last week on June 3 and 4. Other key takeaways: 40% of those CEOs polled expected the U.S. economy to grow; while only 23% held that view back in April. In fact, in June, 51% of those CEOs polled said they expect conditions to continue to improve as trade negotiations settle. The monthly survey first began in 2002, and includes several data points that show how U.S. business leaders view the economy.
