An arbitrage marketer sets up a proxy through a Chrome extension, launches a Facebook Ads account — and after a couple of days receives an IP ban. Everything is correctly configured in the settings: login, password, port. The problem is that the proxy extension does not cover all browser traffic, and sometimes does not touch DNS requests or WebRTC connections at all. We will explore how a proxy in an extension differs from a system proxy, what exactly can leak, and how to check this in 5 minutes without involving tech support.
How a Browser Extension Proxy Works
Extensions like Proxy SwitchyOmega, FoxyProxy, or built-in proxy managers of anti-detect browsers operate at the browser API level chrome.proxy. This means that the proxy is applied only to the traffic that passes through the browser's engine — that is, to the tabs you open within Chrome or a Chromium fork.
It sounds safe, but in practice, this approach has blind spots. The extension does not control system processes, background updates of the browser itself, requests from certain built-in Google services (synchronization, security checks), and does not always intercept WebRTC traffic, which is used in video calls and some advertising pixels to determine the user's real location.
An additional problem is DNS requests. Some extensions only proxy TCP connections to web pages, while DNS resolution (turning a domain into an IP) occurs through the system resolver, bypassing the proxy. As a result, your internet service provider can see which sites you visit, even if the traffic itself goes through a proxy server in another country.
How a System Proxy Works
A system proxy is configured at the operating system level — in the network settings of Windows, macOS, or Linux. In this case, all device traffic passes through the proxy server: browser, email client, messengers, background application updates, and system services.
This approach is more reliable in terms of leaks, but creates another problem for multi-accounting: if you manage 10-20 accounts simultaneously in different anti-detect profiles, the system proxy is one for the entire machine — you cannot assign different IPs to different profiles without using virtual machines or containers. That is why anti-detect browsers like Dolphin Anty, AdsPower, GoLogin, Octo Browser, and Incogniton implement their own proxying mechanism at the profile level, which essentially emulates a system proxy but is isolated for each browser window.
If you are working with one account on one device — a system proxy through Windows settings or a VPN client with proxy mode will be more reliable than a regular extension. But for farming 20-30 accounts, this option is not physically suitable.
Comparison Table: Extension vs System
| Parameter | Proxy in Extension | System Proxy |
|---|---|---|
| Traffic Coverage | Only browser tabs | All device traffic |
| DNS Requests | Often bypass the proxy | Usually through the proxy |
| WebRTC Leak | Requires separate disabling | Partially closed at the network level |
| Different IPs on Different Profiles | Possible (in anti-detect browsers) | No, one IP per system |
| Suitable for Multi-Accounting | Yes, if the browser supports profiles | No without virtualization |
What Can Leak Through a Proxy Extension
Let's analyze specific leak scenarios that arbitrage marketers and SMM specialists encounter in practice.
WebRTC. This technology is used for video calls directly in the browser and can expose your real IP address through a STUN server, even if all other traffic goes through the proxy. Facebook, Google, and TikTok can read WebRTC data through JavaScript on the page, and if WebRTC is enabled without protection — the system sees two different IPs: the proxy and the real one. This is one of the most common triggers for banning advertising accounts.
DNS Requests. Even if HTTP/HTTPS traffic goes through the proxy, domain name resolution may occur through your provider's DNS server. For simple browsing, this is not critical, but some anti-fraud systems of marketplaces and advertising platforms compare the geolocation of the DNS server with the geolocation of the IP — and find discrepancies.
Background OS Processes. Windows updates, iCloud synchronization, messenger notifications — all of this works bypassing the browser extension. For clean farming of social media accounts, this is not always critical, but if you are simultaneously using the desktop application of the same platform (for example, Telegram Desktop alongside the web version), systems may match activity by time and IP.
Extensions and Plugins. Other installed extensions in the same browser — translators, password managers, ad blockers — may make their own network requests bypassing the proxy if they do not operate through the tab API and use system sockets.
Real Cases of Bans Due to Leaks
An SMM agency manages 25 client Instagram accounts through one anti-detect browser, proxies are assigned through the built-in profile manager. After a week, 8 out of 25 accounts receive a shadow ban. Upon checking, it turns out that WebRTC was not disabled in the profile settings, and Instagram saw the real IP of the agency's office across all eight accounts simultaneously — a classic chain ban due to matching a hidden parameter.
Another example — an arbitrage marketer launches Facebook Ads with residential proxies from the USA, but DNS requests go through a local provider in Russia. Facebook detects a mismatch between the system's time zone, DNS geolocation, and the proxy IP — the account gets banned during verification without explanation.
For tasks where the purity of the IP-geolocation-DNS combination is critical, residential proxies usually prove to be more reliable than data center ones — the provider issues the real IP of the end user, which is less likely to end up on anti-fraud systems' blacklists. You can read more about the difference on the residential proxies page.
How to Check IP and DNS Leaks in 5 Minutes
Before launching any advertising account or SMM profile, it is worth conducting a quick check. You do not need technical skills — just open a browser with the proxy set up and visit one of the checking services.
- Open the site browserleaks.com/webrtc in the profile with the proxy — if another IP is visible in the "Local IP Address" or "Public IP Address" section, it means WebRTC is not protected.
- Visit whoer.net — the service will show the final verdict on the match of IP geolocation, DNS, and system time zone. If there is a discrepancy in at least one item — the profile is not ready for work.
- Check the "DNS Leak Test" section on dnsleaktest.com — the service will show which DNS servers your requests go through. If among them is your real internet provider's server — DNS is leaking past the proxy.
- Compare the system's time zone with the proxy's geolocation. If the proxy provides an IP from Germany, and the system is set to Moscow time — this is also a red flag for Facebook and TikTok anti-fraud systems.
Conduct this check every time you create a new profile or change the type of proxy. It takes less than 5 minutes but saves hours on recovering banned accounts.
Setup in Dolphin Anty, AdsPower, Multilogin
Modern anti-detect browsers solve some of the proxy extension problems through a built-in proxying engine at the profile level, rather than an extension. Let's break down the setup using popular tools as examples.
Dolphin Anty: when creating a profile, open the "Proxy" tab, select the type (HTTP, SOCKS5), and enter the IP, port, login, and password. Importantly — immediately after saving, go to "Fingerprint" and check that the WebRTC parameter is set to "Based on IP" or "Disabled," not "Real" — otherwise, the browser will continue to expose the real IP through WebRTC over the proxy.
AdsPower: when creating a profile in the "Proxy" section, select "Custom" and manually enter the proxy data, or connect via API for bulk profile creation. In the "Fingerprint Config" settings, be sure to check the WebRTC item — by default, it may be set to leak the local IP.
Multilogin: the proxy is assigned at the profile level through "Proxy Manager," supporting connection checks directly from the interface with the "Check" button before launching the profile — convenient to avoid wasting time launching with a non-working proxy.
For all three tools, a common recommendation is not to use Chrome proxy extensions within an anti-detect browser if it already has a built-in proxy manager. Double proxying through an extension over the browser's system mechanism often creates the very DNS and WebRTC leaks mentioned above.
For farming Facebook and Instagram accounts through anti-detect browsers, many choose mobile proxies — IPs from mobile operators are less likely to raise suspicion from anti-fraud systems, as one IP usually serves thousands of real mobile internet users.
Checklist Before Launching an Account
- WebRTC is disabled or set to "Based on IP"
- DNS requests go through the same proxy as the main traffic (checked on dnsleaktest.com)
- The system's time zone and the proxy's geolocation match
- The system and browser language corresponds to the proxy region
- Chrome proxy extensions are disabled if using the built-in manager of the anti-detect browser
- Each profile has a separate unique IP, without reusing it on different accounts
- Check on whoer.net or browserleaks.com passed without red marks
Conclusion
A browser extension proxy is convenient for a quick start, but has blind spots — WebRTC, DNS requests, and background system processes can expose the real IP even with the proxy enabled. A system proxy is more reliable in terms of traffic coverage, but is not suitable for multi-accounting without virtualization. The optimal option for arbitrage marketers and SMM specialists is a built-in proxy manager in an anti-detect browser with mandatory checks for WebRTC and DNS before each profile launch.
If you are testing ads or managing multiple accounts on Instagram, TikTok, and Facebook Ads, it is worth choosing proxies that initially reduce the risk of mismatching geolocation and DNS. For most farming and account management tasks, residential proxies are suitable — they provide a high level of anonymity and are less likely to raise suspicion from the platforms' anti-fraud systems.