BADBOX 2.0: The Botnet Living in a Million Living Rooms
The cheapest Android TV box on the shelf may already be working for someone else. In June 2025 the FBI warned that the BADBOX 2.0 botnet has infected millions of home devices, and the infected boxes sell their owners' home internet connections as residential proxy services. Your store's bot problem may be running on a television.
What BADBOX 2.0 is
The FBI issued a public service announcement on June 5, 2025, alert number I-060525-PSA, warning that cyber criminals are exploiting Internet of Things devices on home networks using the BADBOX 2.0 botnet. The compromised devices include cheap Android TV boxes, streaming sticks, digital projectors, digital picture frames, and aftermarket vehicle infotainment systems. The original BADBOX campaign was identified in 2023 and disrupted in 2024. BADBOX 2.0 came back with a wider infection path.
The infection happens before the buyer ever opens the box, or during setup. Criminals configure the product with malicious software before purchase, or infect it when the user downloads required applications from unofficial marketplaces during setup. Once the device connects to the home network, it becomes part of the botnet, and the botnet maintains backdoors to proxy services that criminals exploit by selling or providing free access to compromised home networks.
Google's own action shows the scale. In July 2025 Google filed a federal lawsuit against the botnet's operators, describing it as the largest known botnet of internet-connected TVs, affecting more than 10 million Android devices. HUMAN Security's research, conducted with Trend Micro, put the count above 1 million consumer devices at first disclosure, and Google's filing put it above 10 million. The gap between the two numbers is a measurement difference, not a contradiction: HUMAN counted what it could confirm, Google counted what its telemetry showed.
What the botnet does with those living rooms
The infected devices serve several criminal purposes at once. They commit programmatic ad fraud and click fraud in the background, running fake ad views that look like real users because the traffic comes from a real home IP address. They create fake accounts on social media and banking sites. They conduct account takeover attacks, trying stolen credentials against login pages. And they sell access to the home network through residential proxy services, letting other criminals route their traffic through a residential IP that no blocklist flags. Related reading: How to Tell If Bots Are Skewing Your Store Analytics.
That last purpose is the one that matters most to a store operator. An IP blocklist is useless against traffic from a real home in Ohio, because the IP belongs to a real home in Ohio. The device owner has no idea. The traffic looks human because the network is human. The only signals left are behavioral: sessions with no mouse movement, no scrolling, no engagement, and a speed no human matches.
The FBI's warning signs for buyers are straightforward: devices from unrecognizable brands, marketplaces that require disabling Google Play Protect, and devices that are not Play Protect certified. A streaming box that costs a third of the brand-name equivalent is cheap because someone else is paying the difference, and the payment is the buyer's network.
What this means for a store
Stop trusting IP reputation alone. Residential proxies built on BADBOX devices defeat every IP-based defense, including the data-center IP lists that catch general invalid traffic. The defense that works is behavioral: challenge sessions that move too fast, show no engagement, or repeat the same path at machine speed, and use the signals from your own analytics rather than a vendor's blocklist.
And check your own analytics for the shape of this traffic. Sessions with near-zero engagement time, a single page view, and no interaction, arriving from residential IPs in every geography at once, are the signature. They inflate session counts, distort conversion rates, and poison the A/B tests that depend on clean traffic. A store that does not filter them is optimizing against a ghost audience. More on this: Bots Are Inflating Your Cart Abandonment Rate. Your Funnel Is Lying..
Frequently asked questions
Is my store affected by BADBOX 2.0?
Possibly, if you see residential-IP traffic with no engagement and no interaction. The botnet's devices are real homes, so the traffic looks legitimate to IP-based filters.
How do I defend against residential proxy bots?
Use behavioral signals, not IP lists. Challenge sessions that move at machine speed, show no mouse movement or scrolling, and repeat the same path.
Are the device owners at fault?
No. The malware is installed at the supply-chain level or during setup, and most owners have no idea their device is compromised. The FBI's guidance is aimed at buyers, not at blaming them.
How big is the botnet, 1 million or 10 million?
Both numbers are published. HUMAN Security confirmed above 1 million devices at first disclosure. Google's July 2025 lawsuit filing put it above 10 million based on its telemetry.