U.S. cybersecurity giant CrowdStrike recently warned that Chinese-related hacker organizations are increasing their cyber attacks on U.S. technology companies, focusing on key assets in the field of artificial intelligence to narrow the technological gap with the United States.
CrowdStrike pointed out in a newly released technology threat landscape report that in the past year, entities related to China accounted for more than 58% of state-level targeted cyber attacks targeting technology companies, especially their artificial intelligence assets. The company said in a statement that these "adversaries with Chinese background are escalating espionage activities against technology organizations to steal artificial intelligence capabilities and intellectual property that are difficult to develop independently in a timely manner."
The time range of the report statistics is within one year as of March 31. CrowdStrike believes that the United States’ restrictive measures against China’s acquisition of advanced AI training chips have curbed the development of Beijing’s related technologies to a certain extent. At the same time, China's local artificial intelligence models are trying to provide intelligent performance close to the international leading level while reducing operating costs.
In addition to attacks around artificial intelligence technology, CrowdStrike also pointed out that China-related cyber operations also targeted government communication systems in some Southeast Asian countries and "continuously" maintained access to the networks of some technology organizations in North America by exploiting security vulnerabilities. The Cyberspace Administration of China has not yet responded to CNBC’s faxed request for comment.
Earlier this year, U.S. artificial intelligence companies Anthropic and OpenAI publicly complained that some Chinese companies obtained competitive intelligence from U.S. technology products through "refining" and other technical means, triggering discussions about whether the boundaries of relevant behavior constituted "illegality" or "violation." Some analysts warned at the time that, in the context of the accelerating global AI competition, there is still a large gray area in how to define the compliance boundaries for technical intelligence acquisition.
In recent weeks, Anthropic has been touting the cybersecurity capabilities of its latest Mythos model and has made the technology available to companies such as CrowdStrike. On Tuesday, Anthropic released an open version of the model, "Claude Fable 5," to the public. According to the evaluation of Artificial Analysis, a third-party rating agency, this model's score on the intelligence index is "almost 5 points ahead of the strongest models in other laboratories."
In addition to China-related operations, CrowdStrike also disclosed in the report that North Korea-related hacker groups are also trying to infiltrate IT practitioner groups in North America, Europe and Asia, with the main purpose of raising funds for their regime. These actions are seen as another manifestation of national-level cyber actors using technology and cyberspace to achieve strategic goals in the current complex geopolitical environment.
Full report:
https://www.crowdstrike.com/en-us/blog/crowdstrike-2026-technology-threat-landscape-report/