How Bright Data Turns Smart TVs Into Proxy Nodes for AI Scrapers

Include Security’s reverse engineering of Bright Data’s iOS SDK reveals how the company builds its 400M-IP residential proxy network by embedding code in consumer apps, including connected TV apps, that turns user devices into exit nodes for paying scraping customers. Anti-bot defenses from Cloudflare, DataDome, and HUMAN have pushed AI training pipelines toward residential proxies, and Bright Data positions itself as the legal alternative to botnets like Aisuru and trojanized SDKs like PROXYLIB. Smart TVs are the prize target: always powered, always connected, never battery-constrained, and gated only by a consent dialog navigated with a remote.

The SDK pulls config from an unauthenticated endpoint that anyone can query with an app bundle ID and version string, exposing idle thresholds, VPN-bypass flags, cross-platform identity stitching, per-country bandwidth tiers, and a partner manifest naming integrations like PlayWorks, CloudTV, and Longvision. Petflix’s opt-in screen promises Bright Data will “occasionally” use device resources, but the SDK config sets a default 200 GB monthly WiFi budget. After config fetch, the SDK opens a persistent WebSocket to proxyjs.brdtnet.com, which still serves a TLS cert for luminatinet.com — Bright Data’s pre-2018 name — giving network defenders a clean indicator to distinguish peer-tunnel traffic from legitimate customer-side proxy usage.

The disclosure matters because legal residential-proxy supply has escaped the scrutiny applied to botnets, even though the consent model collapses on a 10-foot interface where users cannot meaningfully evaluate that their home connection is being rented out for commercial scraping. The unauthenticated config and partner endpoints also mean researchers, network operators, and competitors can map the network’s behavior without ever installing the SDK.