The Chinese IA company Deepseek says that it faces “large -scale malicious attacks”, which affect its services.
Chinese state media cite cybersecurity experts who say that attacks come from IP addresses based in the United States, which the BBC is unable to check.
This comes for hours after the White House press secretary Karoline Leavitt said that the United States National Security Council (NSC) “plans” Deepseek.
The little -known Chinese company’s AI model would have surprised Silicon Valley, which made folies on billions on the IA infrastructure – last week Trump announced an AI plan worth half a bill of dollars involving the best American companies.
Deepseek did not immediately respond to a request for comments from BBC News.
“Due to large -scale malicious attacks against Deepseek services, registration can be occupied,” a banner said on the company’s website on Wednesday.
This is the second attack of this type that Deepseek reported this week – but Yuyuan Tantian, a chain of social media from the Chinese state diffuser CCTV, said that the company had to face “several” cyberattacs last weeks, which have increased in “intensity”.
Deepseek drew towards glory that last week when the Geeks AI praised his latest AI model and people started downloading his chatbot on application stores. Its increase has caused a collapse of American technological actions, many of which have since recovered land.
But the IA industry of America was shaken by the apparent breakthrough, in particular because of the dominant opinion that the United States was far in the race. It is believed that a series of commercial restrictions prohibiting China access to high -end fleas would have cemented this.
Although China has strengthened investments in advanced technologies to diversify its economy, Deepseek is not one of the large Chinese companies that have developed AI models to compete with the American manufacturing chatgpt.
Experts say that the United States always has an advantage – it houses some of the largest tokens manufacturers – and that it is not yet exactly how Deepseek has built its model and how far it can go.
But the White House has raised national security problems in the midst of reports that the US Navy prohibited its staff from using the Deepseek application.
“I spoke with [the National Security Council] This morning, they look at this [natinal security implications] Perhaps being, “said Leavitt.
Speaking on Fox News, the “AI of the White House recently named and Czar”, David Sacks, suggested that Deepseek may have used the models developed by the American company OpenAi to improve.
This process – which implies a model of learning AI of another – is called distillation of knowledge.
“There is substantial evidence that what Deepseek has done here is that they have distilled the knowledge of Openai models,” said Sacks. “I think that one of the things you will see in the coming months is that our main AI companies take measures to try to prevent distillation … which would certainly slow down some of these copying models.”
Openai has echoed this in a later declaration that the Chinese and other companies “constantly try to distill the models of leaders of American AI companies”.
“As the first manufacturer of AI, we are engaging in countermeasures to protect our [intellectual property]… and believe that we are advancing that it is extremely important that we are working closely with the American government to best protect the most capable models. “
While Deepseek rocked the markets this week, US President Donald Trump described it as “an alarm clock” for the American technology industry, while suggesting that it could ultimately prove to be a “positive” sign.
“If you could do it cheaper, if you could do it [for] less [and] Access the same final result. I think it’s a good thing for us, “he told journalists on the Air Force One.
He also said that he was not concerned about the breakthrough, adding that the United States will remain a dominant actor on the ground.
Fan Wang additional reports
