Let's start at the beginning.

On October 1st of 2025, Walmart, in partnership with Google, released the first of their affordable onn brand home security hardware. This included the onn Indoor Camera, and the onn Video Doorbell. These new products were the first of the Google Home Gemini built in Program. Which gives third party vendors access to Google's own smart home software, APIs, and reference designs. This includes smart cameras, and starting in 2026, smart speakers. This year, Walmart and Google expanded on this partnership by introducing the onn Outdoor Camera and onn Outdoor Floodlight Camera.

It's been known that Google's home security tech is, to say the least, overpriced. Which is why onn's new offerings are so attractive. I've compiled a table comparing Google's Nest hardware to all currently available onn equivalents.

Google also offers battery powered variants of some of these devices, albeit at an increased price, and lacking continuous video recording.

The numbers don't lie.... these are kind of a steal. So, naturally, I picked a few up. The doorbell, a few outdoor cams, and a floodlight cam.

Where it went wrong

Setup and installation went without a hitch, notably using the Matter protocol, a first for devices of this kind. My concerns arose when Familiar Face Detection was noticeably missing from the setup process.

Familiar Face Detection, or "Familiar Faces", is a facial recognition tech that learns the faces of people who visit your home over time. It's a standout feature of Google's smart home hardware, and their Google Home Premium subscription. It is also a feature explicitly advertised on the product listing, and official launch blog posts for these exact onn. outdoor cameras. Evidence below.

Image courtesy of walmart.com

Of course, my first assumption was, "this can't be right, I must have broke something". So, I factory reset the cameras, to no luck. Factory resetting a second time didn't yield better results. Which lead to my next logical step, reaching out for assistance.

Where Customer Support Goes to Die

I should have known that wouldn't go well. I can't name the last time I've had a pleasant experience speaking with customer service agents. It's not their fault, honestly, they just aren't given the right tools, resources, or peers to effectively do their jobs. Nonetheless, I tried anyway.

After some back and fourth with an agent, I was told the cameras lacked the feature entirely.

After elaboration and explanation that the cameras are supposed to have the feature, I unfortunately hit a brick wall, and was directed towards Walmart support.

This made ZERO sense, especially considering the cameras run GOOGLE software, but like the agent suggested, I reached out to Walmart. First, by email, which the support address of, could only be found in review replies on the Walmart website.

This email goes completely ignored. So next, I call their support number. This number was found in physical product manual. The agent didn't even know about the product itself, and was zero help. They assumed it was a problem with Google.

A month of reaching out to the two companies goes by, with absolutely zero progress. These are trillion dollar mega corporations, by the way.

Next, I left a review on the product page at walmart.com explaining my issue, hoping it could reach the proper people that way. This review remained in the "pending" state for a week or so, and never got approved. When all hope was lost, I remembered one crucial detail.

I'm literally a programmer.

I instantly got to work picking apart at the cameras and finding out anything I could about them. This gets technical, so, feel free to skip ahead.

• Interestingly, the Matter protocol is only used for activating the camera's built-in lights.
• The cameras open zero ports over LAN.
• They appear to be manufactured by a company called ITON Technology Corp. (don't quote me on that)
• They run a variant of the same firmware Google's Nest Cameras/Doorbells use, even using the same code branch (OPENMASTER). This software is very locked down.
• According to Google, onn cameras store Familiar Face data in the cloud, versus locally on Google Nest cameras. This eliminates the possibility of it missing due to a hardware limitation.
• The flaw is limited to the security cameras specifically;

As I mentioned, I also obtained an onn Video Doorbell. Interestingly, Familiar Face detection runs flawlessly on that device. So I dug deeper.

Pulling camera configurations with Google Takeout, a feature allowing Google Account owners to request access to their data, I was able to understand how they work on Google's servers.

Familiar Face library

Notably, all devices share the same Familiar Face library on Google's servers, which resides inside a face_library directory. That contains a JSON file, categorizing images stored, and nested directories containing captured images of each identified face, labeled with that individual's UUID.

Here's an anonymized example of the JSON file, face_library.json

[
  {
    "id": "12345678-abcd-1234-efgh-1234567890ab",
    "category": "HUMAN_CATEGORY_KNOWN",
    "hero_exemplar_id": "9876543210987654321",
    "human_exemplar_info": [
      {
        "id": "9876543210987654321",
        "detection_time": {
          "seconds": 1600000000,
          "nanos": 123456000
        },
        "event_session_id": "1600000000",
        "track_id": "1600000000000000-123456789",
        "face_net_signature": "eW91J3JlIHF1aXRlIG5vc2V5LCBsb2w="
      },
      {
        "id": "9876543210987654322",
        "detection_time": {
          "seconds": 1600000500,
          "nanos": 987654000
        },
        "event_session_id": "1600000450",
        "track_id": "1600000450000000-987654321",
        "face_net_signature": "bm8gZmFjaWFsIGRhdGEgZm9yIHlvdQ=="
      }
    ],
    "name": "Logan"
  }
]

There are a few notable bits of data here.

• The capture time includes nanoseconds??
• category can either be HUMAN_CATEGORY_KNOWN or HUMAN_CATEGORY_NOT_A_HUMAN. This is probably for Animal detection, another feature I'm unable to use on my onn Outdoor Cameras. For now.
• Facial mapping data is hashed into a multi-dimensional array of floating point numbers, a one-way facial vector array (embedding), and then encoded into Base64 stored in face_net_signature. Very efficient.

Camera configurations

Next, we have the cameras themselves, which get their own config files. Several. Too many to list. If you're curious, shoot me an email with any questions you have, or try poking at your own Google Home cameras using Google Takeout.

The file we're interested in, is nest.trait.product.camera.CameraEntitlementTrait.json. Which dictates which features the cameras can and can't use.

Here's the onn Video Doorbell's entitlement config.

{
  "cvr_hours": 240,
  "ebr_hours": 1440,
  "are_zones_allowed": true,
  "hours_of_free_tier_history": 3,
  "vision_trigger_entitlements": {
    "motion": {
      "enabled": true
    },
    "person": {
      "enabled": true
    },
    "face": {
      "enabled": true
    },
    "vehicle": {
      "enabled": true
    },
    "pet": {
      "enabled": true
    },
    "package": {
      "enabled": true
    },
    "garage_door": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    }
  },
  "sound_trigger_entitlements": {
    "sound": {
      "enabled": true
    },
    "person_talking": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "dog_barking": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "glass_breaking": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "smoke_alarming": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "co_alarming": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    }
  },
  "door_trigger_entitlements": {
    "garage_door": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    }
  },
  "video_analysis_trigger_entitlements": {
    "frame_embedding_generation": {
      "enabled": true
    }
  },
  "etr_video_preview_seconds": 10
}

Here's the onn Outdoor Camera's entitlement config.

{
  "cvr_hours": 240,
  "ebr_hours": 1440,
  "are_zones_allowed": true,
  "hours_of_free_tier_history": 3,
  "vision_trigger_entitlements": {
    "motion": {
      "enabled": true
    },
    "person": {
      "enabled": true
    },
    "face": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "vehicle": {
      "enabled": true
    },
    "pet": {
      "enabled": true
    },
    "package": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "garage_door": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    }
  },
  "sound_trigger_entitlements": {
    "sound": {
      "enabled": true
    },
    "person_talking": {
      "enabled": true
    },
    "dog_barking": {
      "enabled": true
    },
    "glass_breaking": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "smoke_alarming": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    },
    "co_alarming": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    }
  },
  "door_trigger_entitlements": {
    "garage_door": {
      "reason": "DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE"
    }
  },
  "video_analysis_trigger_entitlements": {
    "frame_embedding_generation": {
      "enabled": true
    }
  },
  "etr_video_preview_seconds": 10
}

Most of this is self-explanatory.

• door_trigger_entitlements is garage door detection, which a Redditor noted the indoor camera doesn't support. Despite running the same software. Just like Familiar Face Detection! I'm noticing a pattern here.
• sound_trigger_entitlements is auditory triggers, like dogs barking, people talking, things you might appreciate your camera being able to detect.
• video_analysis_trigger_entitlements is used for Gemini for Home. Which uses Google's Gemini AI models to generate rich descriptions of video events. I.e. "animal seen" v.s. "a dog rolls around in the grass". Paired with Familiar Face Detection this is a very powerful tool. If only.

And, the reason I went through all this trouble, "face", which is not enabled, with the reason DISABLED_REASON_ENTITLED_BUT_DEVICE_NOT_CAPABLE.

Breaking this down is simple. The feature is DISABLED, I am ENTITLED to the feature, but, the DEVICE is NOT_CAPABLE.

Funny.

It's not like it's advertised on the store listing, Google support articles, and Google blog posts. (Sarcasm)

Are we serious

So, let's lay out the facts.

• The cameras should have the feature.
• They don't.
• Support is USELESS.
• Neither Walmart nor Google wanted to take accountability.
• Walmart doesn't want customers to find out the feature is missing.

What on EARTH do I do next?

I state my case publicly on Google's own community forum for Google Home. A few days go by, zero replies. Then, they delete my post.

Stating my question "violates the Community Policy". Sorry for trying to save them from a PR disaster, I guess.

I'm pissed, at this point. So, I take it to Reddit.

My smoking gun

My Reddit post does not fail in the slightest.

Within hours, the community rallied behind my hard evidence. People decide against purchasing the hardware, others share their stories and struggles with the hardware, people wanted ANSWERS. It's easy for corporations to dismiss a complaint as "user error," but it's much harder to ignore raw JSON logs that explicitly expose what's wrong, and public outcry.

The public pressure worked. Shortly after the post gained traction, the official Google Nest Community team was forced to break their silence and reply directly to my thread.

This is your average PR speak. Let's translate:

Well done, Google.

Because they chose to play corporate hot potato and silence me, they turned a minor launch bug into a massive public embarrassment. And it only gets better from here.

My undeniable proof caught the eyes of the tech press.

Stephen Schenck over at Android Authority published a front-page article exposing the entire debacle linking directly to my Reddit post. Soon after 9to5Google's Ben Schoon published their take, and even Ryan Merket of RuntimeWire reported on it.

And, would you look at that. After the dust settled, the onn team finally replies to my email!

Thank you, onn support. Only took a month, and a PR disaster I was pissed enough to cause.

Others spoke out

One Redditor reveals they reached out to onn support as well, even getting a response from them, mentioning onn was investigating the issue. Not only did onn know, but they kept it on the down low, hoping nobody else would notice.

My community post gets reinstated

Finally, my post on the Google Home and Nest community gets reinstated. A community specialist (their word for moderator) even mentioning how my blog post (you're currently reading it) is what helped them find it to resolve the issue. No hard feelings.

To put it bluntly

Massive tech companies only move as fast as public pressure forces them to.

Had I accepted the endless cycle of Tier-1 support scripts and given up when my forum post was deleted, onn would still be ignoring my emails, and these cameras would still be on store shelves with unsuspecting customers not aware of the broken launch feature, while still footing the bill.

​And, most importantly, I'd still be without Familiar Face Detection.

They should offer me a job at this point.

Don't let tech big tech gaslight you into thinking it's user error. Check your logs, save your receipts, and hold them accountable.

Fair Use Notice: This post contains commentary and criticism of commercial products and media. All trademarks, logos, and third-party media layouts remain the property of their respective copyright owners.