376
Android users can't remove nude-scanning app as Google opens the feature to app developers (cybernews.com)
submitted 3 weeks ago by to c/technology@lemmy.world
111 comments fedilink hide all child comments
 

cross-posted from : https://lemmy.zip/post/70348978

you are viewing a single comment's thread
view the rest of the comments
[–] 8 points 3 weeks ago (3 children)

It’s not looking for stuff like you’re describing, skeletons, graves, birds, whatever spoopy “dark” stuff you put on there. It’s looking for nude people, then actioning/alerting if it thinks minors are in the images. You definitely don’t want to honeypot this ai bro.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 1 point 3 weeks ago (2 children)

    That's the thing: they're looking for something, and in order to look for something, they must look, and lots of computation (AI inference) must happen, especially when it involves nonindexed files (my own private art, from digital drawings and handmade sketches, going through photos of my own ritualistic installations, all the way to Blender 3D renders and animations, all of which is yet to be crawled). And if this involves gibibytes of data, on a smartphone whose hardware can barely run 1B parameter clankers locally without hiccuping, I can see my private files going to Google's servers for "efficient inference", files of which will likely end up seen by (and sowing some interesting inspirations for nightmares in the subconsciousnesses of) Google's employees during server maintenance/audits or, mainly, end up training their clanker.

    As for the latter, hopefully Google Gemini would end up "seeing" enough of my digitized psyche to randomly start silly infodumpings about, for example, different owl taxonomies and their diets down to the very gory microdetails about how owls make a tasty dinner out of mice, and/or the different interpretations by historians and archaeologists regarding the figure (Ereshkigal or Lilith) depicted in the Burney Relief, a Mesopotamic plaque of which, by the way, is part of the phone's gallery too (including my own artistic works inspired on it) and seemingly an efficient trigger for nudity detection algorithms, which will likely put the rest of the gallery under automated scrutiny (because there was a positive, albeit a false one) and, thus, serving as the semantic Trojan Horse for my dark artistic private content to make its way to their servers.

    I mean, how else Google Gemini would've taken the "inspiration" to suddenly start telling people to "eat rocks" and "add glue to pizza"? Its training data is crowdsourced so there's nothing much Google can do to avoid sources poisoning the well (and I must nod at the Nephentes algorithm which I praise as way more efficient than Anubis to hurt the western AI corps) other than:

    1. Adding said poisoned sources to their .gitignore equivalent if they know they'll likely find poison they don't want messing up with their expensive AI training.
    2. Doing manual curation of people's personal files, which is time-consuming and potentially unfeasible when dealing with billions of phones.
    3. Trying to improve the detection algorithms to avoid poisoned sources (which requires training data, which requires them to rely on potentially poisoned sources, which defeats the whole purpose of improving the algorithms in the first place).
    4. Stopping to put everything else into scrutiny when the detection algorithms stumble upon (potentially false) positives, risking false negatives, which will likely attract some heat from regulatory agencies worldwide and fines for failing to comply with things such as age checking laws (and there it goes the GOOG stocks, together with the entirety of NASDAQ, like a Titanic as investors realize they're paying for fines not theirs; be it GOOG or other stock, the popping of the ongoing AI bubble will inevitably be something in this regard, Western corporations having their clankers failing on them).

    At best, this will consume some portion of Google's resources (including but not limited to human resources), albeit as infinitesimal as my silly existence can get to a worldwide corporation, but it's still something I guess...

    !technology@lemmy.world

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 5 points 3 weeks ago (1 child)

    Image classification is old tech that existed long before ChatGPT and similar huge models became the new norm. These models are already embedded in many data flows and run within milliseconds on non-specialized hardware.

    You were talking about 1B models. More modern image classification are in the mere millions. Yolo v10 has like 7M, MobileNet, which Google often uses, has around 4M parameters depending on the version. Image classification is not why there are data centers being built everywhere.

    Camera apps, image and video filters and content moderation (the app in question) all use them for live processing. I can guarantee you this runs flawlessly on your phone unless it's older than 10 years.

    You're employing AI guerilla warfare against pre-LLM technology. It's not effective.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 1 point 3 weeks ago

    Machine translation is also pre-ChatGPT tech, yet Google have been outsourcing their Translate to their Gemini, as in, Google Translate became yet another Gemini interface, just like many Google "products" nowadays.

    But I'm aware of and I agree that "protoclankers" such as Image Classifier models (e.g. MobileNet) are lightweight enough to run on low-end devices like the one (Nokia 5.4) I use... although, even in this case, when protoclankers are employed in place of clankers, one of my main points stays the same: the steps being performed whenever there's a (false) positive for a pre-instructed pattern.

    A false positive is particularly likely for protoclankers due to their higher compression. If even full-fledged clankers such as Mistral can misread my owl drawings as "a cat" or (because we're talking about Google) Google Lens/Gemini can misread my digital art of a solar eclipse silhouetted by a Bubo virginianus as "a viral photo showing a Spanish professional skateboarder Danny León performing a mid-air trick perfectly aligned with the August 12, 2026 total solar eclipse" (when it's far from being a photo), I can only imagine how awfully a million-parameter Image Classifier protoclanker can perform with deeply-symbolic and carefully-drawn art which even humans can struggle to understand.

    There's no magic when it comes to how much data can fit within a mathematical representation of a dataset larger than the representation itself, especially when Zipf's law gets applied to said dataset in order to try and remove hapax legomena while also accounting for repeated data.

    And what happens after a false positive happens? Even if it's a locally-ran image classifier, the whole thing isn't fully local; if the locally-running protoclanker detects something matching the pre-instructed pattern, then a set of online tasks will likely be performed: uploading the file(s) to Google servers or, if the pattern being searched for is criminal imagery, preemptively filing a report to the authorities.

    In the latter case, if the protoclanker is filing reports without humans in the loop doing the due-diligence of double-checking (which is very likely what's happening due to how employees cost money and how Google et al have been prioritizing their budget to the ongoing AI hype), then Google, legally responsible for their protoclanker, would be committing the crime of false accusation (e.g. misreading my derivative mashup between Burney Relief and Éliphas Levi's Baphomet as "CSAM" as it incorrectly labels the flanking owls as "minors").

    And given how my artistic content is anything but criminal (the most shady aspects being the psychological horror and the depictions of blood in vampire contexts, aspects of which, given how there are Hollywoodian horror movies depicting the same, are perfectly legal), stemmed from a legitimate religious expression which (at least on paper) is protected by law (Article 5, item VI from Brazilian Federal Constitution as I'm someone under the umbrella of Brazilian laws) and done in a private space, the potentially unsupervised filings of false criminal accusations by Google's protoclanker, be it against me or any other law-abiding person, will hopefully result in Google shooting themselves in the foot.

    The guerrilla tactic here is me being bold enough not to care about the consequences of using a surveillance device (a phone running stock Android) to register, record and manifest my faith in a very verbose and unfiltered, artistic manner, fully aware that the phone will phone home (pun intended; I don't even bother to use VPNs and PiHole, I'm raw-doggin their surveillance when I'm not on my Linux laptop); and in the meantime, hoping Google will hoard fragments of my dark psyché to train their Gemini (thus poisoning the well in the process; it would be fun to see Google Search/Gemini, out of nowhere, randomly spitting invocation texts written in Akkadian/Sumerian cuneiform, infodumping owl taxonomy and their diet, spitting random ramblings against the Demiurge, quoting pessimist philosophers such as Phillipp Mainlander, and/or having Nano Banana to produce images with glitched artifacts resembling the Seal of Lilith).

    !technology@lemmy.world

  • permalink
  • fedilink
  • source
  • parent