The era of affordable Chinese AI models may be coming to an end as the government considers implementing restrictions akin to those in the United States.
The bargain section of the AI market could soon be shut down by China.
DeepSeek’s R1 sparked global interest in affordable Chinese AI, leading to the development of increasingly advanced systems from Alibaba, ByteDance, and Z.ai. Some of these models can be downloaded, customized, and hosted independently, providing developers with a cost-effective alternative to expensive US platforms.
However, this budget-friendly option may soon be impacted by geopolitical factors, this time due to China rather than the US. According to Reuters, Chinese authorities have discussed with Alibaba, ByteDance, and Z.ai the possibility of restricting international access to the nation’s most sophisticated AI systems. These discussions apparently included closed models, open-weight releases, and technologies not yet publicly available.
The global market could lose one of its most affordable AI options.
Chinese models have made strides internationally thanks to their strong performance and competitive pricing. A prime example is Z.ai’s GLM-5.2, which surpasses American systems in capabilities while being much more affordable. Similarly, Alibaba’s Qwen family has become one of the most popular model lineups in China.
Thus, any restrictions on upcoming releases could eliminate a key source of competition. Companies using Chinese APIs or intending to self-host open-weight models may have to switch to pricier alternatives from OpenAI, Google, Anthropic, or other vendors.
It’s important to highlight that this hasn't been confirmed yet, as Beijing has not reached a decision on the issue. Authorities might restrict API access, block downloads of overseas model weights, impose licensing requirements, or reserve certain leading-edge systems for domestic use. Reuters was unable to ascertain how any final regulations would function.
Models that have already been downloaded would be challenging to retract. The uncertainty mainly revolves around the next generation of systems that developers expect will be made available globally.
China is taking cues from Washington’s approach.
The treatment of advanced models as strategic assets, akin to chips, reflects a change in governmental perspectives on AI. Recently, the US imposed restrictions on foreign access to Anthropic’s leading Fable and Mythos models due to national security concerns. Some of these restrictions on the consumer-oriented Fable system were later eased following the implementation of new safeguards, while access to the cybersecurity-focused Mythos remains restricted to select trusted American entities.
Chinese officials are reportedly contemplating stringent penalties for unauthorized disclosures or theft of proprietary AI technology. Furthermore, discussions have included potential limitations on who can financially support domestic AI startups, thereby tightening Beijing’s control over both the models and the firms that develop them.
---
A Windows 11 bug might be silently consuming hundreds of gigabytes of your storage.
Microsoft has issued a fix for a storage-hogging bug in Windows 11. If your Windows 11 machine suddenly appears to be low on storage, the issue may not stem from your downloads folder or game library. According to Windows Latest, a bug associated with a Windows system file can stealthily eat away tens or even hundreds of gigabytes on the system drive.
The problematic file is called CapabilityAccessManager.db-wal, located within Windows’ Capability Access Manager folder. Reports suggest that this issue may manifest as an unusually high “System files” usage in Windows 11’s storage analysis, despite the Settings app not explicitly identifying the specific file responsible. In some cases, users have reported it ballooning to 200GB or more.
Your next Teams meeting might feature an AI assistant that responds on your behalf.
Microsoft Teams is undergoing a significant update that revamps nearly every aspect of the app, ranging from AI-enhanced meetings to an improved chat interface. Most of the new features are currently in testing, with several slated to launch by the end of summer.
The most intriguing addition is an upgraded AI Facilitator capable of listening in on your meeting, recognizing when someone appears confused, and generating an appropriate response.
A hacker's arrest has revealed how Microsoft can monitor your Windows device.
Microsoft was aware of the websites a teenager visited on his Windows PC, even though he allegedly used a VPN to hide his activities while hacking a US jewelry retailer. Court documents unsealed in the case against Peter Stokes, a 19-year-old dual citizen of the US and Estonia, show that Microsoft provided the FBI with information linked to a tracking tool known as the Global Device Identifier, or GDID.
Other articles
The era of affordable Chinese AI models may be coming to an end as the government considers implementing restrictions akin to those in the United States.
Chinese authorities have allegedly talked about restricting foreign access to the nation's most sophisticated closed and open-weight AI models due to concerns over national security.
