DeepSeek reports campaign of ‘large-scale malicious attacks’ after becoming most downloaded app

“Due to large-scale malicious attacks on DeepSeek’s services, registration may be busy.” With this message, the Chinese artificial intelligence platform that has set off alarm bells among U.S. technology companies warns new users of difficulties in incorporating them into its mobile application or the access page on the internet. Since yesterday, when DeepSeek became the application with the highest number of downloads, it has suffered continuous failures in its registration links. Experts attribute this to various strategies but have not yet identified their origin. Several companies have banned DeepSeek downloads for security reasons.

New user registration requires an email address, a secure password, and a verification code that must be sent by the platform within 60 seconds. However, the code either doesn’t arrive, or does so after that timeframe has expired. Users who are already registered can use the company’s services without any complications.

The company justifies this measure by pointing to a large-scale campaign of attacks, of unknown origin. For Marc Rivero, senior security researcher at Kaspersky, “the term ‘large-scale malicious attacks’ can refer to different types of cyberthreats. In this case, since DeepSeek has limited the registration of new users, it is possible that we are facing a distributed denial of service (DDoS) attack, in which servers are saturated with massive traffic to disrupt their operation. However, it could also be another type of attack, such as massive attempts at unauthorized access or the exploitation of vulnerabilities. Until DeepSeek provides more technical details, it is difficult to confirm the exact nature of the incident,” he explains.

The attacks came after the free artificial intelligence app, similar to popular apps like ChatGPT and Gemini, became the most downloaded app on both Apple and Google’s stores.

Its founders, originally from Hangzhou (China), created it in 2023 and launched it just one year later with less expensive requirements than the now-common applications. The emergence of DeepSeek has caused the shares of the tech companies involved in these developments and their necessary equipment, such as Nvidia processors, to plummet on the stock market.

The application of low-cost artificial intelligence has made both companies and the administration of U.S. President Donald Trump nervous. The Republican magnate announced an alliance of OpenAI, Oracle, and SoftBank to invest up to $500 billion in artificial intelligence shortly after assuming office.

Vivek Ramaswamy, founder of Roviant Sciences and the partner of Elon Musk, owner of X and Tesla, in the newly created Department of Government Efficiency, has called for calm after the esteemed guru of technological investment and creator of Netscape, Marc Andreessen, described the emergence of DeepSeek as a “Sputnik moment” on X, alluding to the first artificial satellite sent into outer space by the Soviet Union in 1957, during the Cold War, which accelerated the space race.

Ramaswamy responded on the same network: “Sputnik-like moments are a good thing. We don’t need to freak out, we just need to wake up.” Trump backed this line by considering the emergence of Chinese AI as a “wake-up call,” urging the U.S. industry to be “laser-focused on competing to win.”

Javier Aguilera, general manager of Ikusi Spain, an advanced technology services company specializing in digitalization and cybersecurity, attributes the large number of downloads to the Chinese market itself and also to government strategies. “In the end, there are underlying political issues that make the needle move more in one direction or another depending on what interests them,” he notes.

Aguilera highlights DeepSeek’s “level of refinement,” but does not consider it a groundbreaking technology: “ChatGPT 03 was already open and available. Anyone on the market can do it. You can take something that is already trained and perfect it. With the subsidies from the Chinese government, it can be done at low cost.”

The expert is skeptical that the current situation will continue. “We are in a complete bubble and everything will go back to normal,” he says, before warning that the technology “cannot be used blithely.”

In this regard, the director of Ikusi Spain warns: “We know how they play and how they use the information that is uploaded to these engines when the [Chinese] government is behind them. We have clients who are blocking their use because they are afraid that they will glean information. We have no evidence, but we do have experience with applications such as Tiktok or with Chinese equipment supply companies. It‘s a question of trust and, as a precaution, organizations are going to go for what’s safest, secure technologies,” he concludes.

In this regard, Ericsson has announced the creation of Cognitive Labs to boost open source AI research in advanced technologies such as Graph Neural Networks (GNN), Active Learning, and Large Language Models (LLM). These are technologies that the company is using for the next generation of mobile communications and will have operational centers with virtual access in Madrid, Málaga and Cairo.

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition