Chinese chatbot DeepSeek has raised concerns in the U.S. over its implications for data and national security, and caught the attention of many users outside of China for appearing to censor itself when asked questions sensitive to the Chinese government.
Category
🗞
NewsTranscript
00:00Dethroning ChatGPT, a new chatbot created by Chinese startup DeepSeek, has become Apple's
00:05U.S. app store's most popular free app.
00:08It's a development that's rocked the tech world.
00:10The company says the chatbot was launched at a fraction of the cost and with less advanced
00:14microchips than those developed by industry giants like OpenAI and Google.
00:19But many users have noticed something else.
00:21The chatbot also appears to censor itself on sensitive Chinese issues.
00:25When asked, who is Ai Weiwei, an exiled Chinese artist and activist who has been openly critical
00:30of the Chinese government, DeepSeek did generate a response, but its answer disappeared moments
00:35later, replaced with an error message in request to change the subject.
00:39This isn't surprising, given China's tight control over its internet and social media.
00:44But it is worrying to many, especially if China takes the lead in AI, with its censorship
00:48laws embedded in its models.
00:50Chinese companies, even if they are relatively very independent, they're still within that
00:56legal structure, within that political system, and held, they're bound to expectations that
01:04go along with that, that it's hard for them to escape expectations of censorship, especially
01:11around the central topics that the Chinese government is not in favor of.
01:18Users are concerned about the risks of DeepSeek's release of its AI model as open source, meaning
01:23users can access its underlying code and modify it for their own purposes.
01:27And so almost overnight, every AI developer in the world got access to kind of like PhD
01:34level intelligence that they could run at home.
01:38This makes it very hard to control what the technology is used for, and it's unclear if
01:41DeepSeek has a safety research team like USAI companies.
01:45Other experts have warned that DeepSeek users' data, which its privacy policy says is stored
01:49within China, could be used by the Chinese government, echoing a concern U.S. lawmakers
01:54have expressed about TikTok.
01:56But others say similar privacy issues exist within USAI products as well.
02:01U.S. President Donald Trump's press secretary has also said the National Security Council
02:05is looking into the model's national security implications, and the U.S. Navy has banned
02:09its personnel from using it.
02:11At the same time, in his confirmation hearing, Trump's Commerce Secretary nominee accused
02:16DeepSeek of stealing from the U.S. to launch its new model.
02:19How could it be more clear than this week when DeepSeek, a Chinese AI, says they were
02:26able to create things dirt cheap?
02:28How?
02:29By leveraging what they've taken from us, stolen from us, or leveraged from us.
02:33It's outrageous and it needs to be addressed.
02:37Lutnick says he wants to put more restrictions on China, adding to existing export controls
02:41that limit China's access to advanced chips.
02:44But security concerns are not stopping users from testing out the new chatbot, which many
02:48say works as well as or even better than ChatGPT.
02:52John Su and Cadence Caranta for Taiwan Plus.