The main topic of these two days is the tightening of internet restrictions in Russia. Providers have started Cheburnet encrypted DNS protocols DoH and DoT, making it harder to bypass blocks and reducing privacy. At the same time, from September 1, laws on AI regulation, the digital ruble, and cryptocurrencies came into force, also affecting digital life.

**DNS Blocks: What's Happening?**

Users of Rostelecom, Dom.ru, Tattelecom, SkyNet, and Beeline have found that encrypted DNS services from Google and Cloudflare stopped working. DoT (port 853) is blocked simply by port number, while DoH (port 443) is detected by SNI in the TLS handshake and then 'black-holed' — the connection hangs. Meanwhile, regular DNS over UDP continues to work, but responses are spoofed: for blocked sites, a 'domain does not exist' is returned. This looks like a centralized update of filtering systems.

**What Should Users Do?**

If you use DoH/DoT, consider switching to alternative methods, such as VPNs with proprietary protocols. Also, it's useful to check if your DNS is being spoofed: you can use dig commands with flags to see who actually responds to queries.

**New Laws from September 1**

Laws on AI, the digital ruble, and cryptocurrencies have come into force. AI developments must now align with 'spiritual and moral values', the digital ruble is available in major banks, and cryptocurrencies can only be bought through licensed market participants, but still cannot be used for payments. Also, mobile numbers can now be kept when moving to another region.

**Privacy and Security**

Android 17 will include support for Encrypted Client Hello (ECH) at the platform level, hiding visited site domains from providers. However, ECH only works if the server supports it, and does not hide IP addresses.

Also, a data leak from charitable projects 'Davayte' and 'Ty ne odin' was reported due to attacks on Stripe. Hackers accessed donors' emails and last four digits of cards. This is a reminder to be cautious with online donations.

**How to Protect Data When Using AI?**

Experts advise disabling data sharing for model training, masking confidential information, deleting old chats, and using local models, e.g., via Ollama. This reduces leak risks.

**Summary**

The period was marked by increased control and new privacy challenges. Users should pay attention to device and service settings, use reliable tools for data protection, and keep up with legislative changes.