Frog Lover

I first messed around with the available options, but then I asked it about something obviously concerning, saying I had a gun and was going to shoot myself. It responded... Poorly. Imagine the vibes of trying to cancel Comcast, when you're suicidal.

Anyway, I tried again to ask for help about something else that would be concerning enough for any responsible company to flag. School was one of their main options, which seems irresponsible - do you really think a child in crisis would read that contract?

I told it about a teacher at school trying to "be my boyfriend", and it immediately suggested I help someone else while I wait for help. I was honestly concerned that it wasn't flagged before connecting. Especially when I realized it was connecting me to children.

I first got someone who seemed to be a child in an abusive home. (Censored for their privacy.) I declined to talk to them because despite being an adult and in an OK mental place - I knew I'm not equipped to counsel a kid through that. If my act of being another kid in crisis was real? Holy shit.

Remember- if my BS was true, that kid would be being "helped" by an actively suicidal kid who's also being groomed by a teacher. Their pipeline for "helpers" is the same group of people looking for help.

I skipped a number of messages, and they mostly seemed to be written by children and young adults with nowhere else to turn. Plus one scary one from an adult whose "problem" was worrying that they'd been inappropriate with a female student, asking her to pull her skirt down "a little" in front of the class. Koko paired this person with someone reporting that they were a child being groomed by a teacher. Extremely dangerous, and if this was an episode of Black Mirror? I'd say it was a little too on the nose to be believable.

I also didn't get the option to get help without being asked... Er... Harassed... to help others. If I declined, I'd get the next request for help, and the next. If I ignored it, I got spammed by the "We lost you there!" messages, asking if I'd like to pick up where I left off, seeing others' often triggering messages while waiting for help, including seriously homophobic shit. I was going into this as an experiment, starting from a good mental place, and being an adult with coping skills from an actual therapist, and I still felt triggered by a lot of what I read. I can't imagine the experience someone actually in crisis would be having.

My message was starting to feel mild in comparison to what some people were sharing - but despite that I was feeling very uneasy about my message being shown to children. There didn't seem to be a way to take it back either.

Then I got a reply about my issue. It was very kind and well meaning, but VERY horrifying. Because it seemed to be written by a child, or someone too young to understand that "Do have feelings for the teacher who's grooming you? If you don't, you should go talk to him." Is probably THE most dangerous advice possible.

Not judging the author - I get the impression they're probably a child seeking help themselves and honestly feel horribly guilty my BS got sent to a young person and they wanted to reply. Because WTF. No kid should be in that position to answer my fucked up question or any of the others like it.

---

Anyway, what can you do if this concerns you, or you've had a difficult experience on Koko, with no support from them or Tumblr?

To reach Tumblr, who officially partners with Koko?

Send a message to Tumblr Support describing your concerns with their partnership with Kokobot

Report kokobot to Tumblr's abuse hotline describing your experience with KokoBot, especially if you are a minor who suffered harm, as they have a legal responsibility to address that.

To get Koko's attention:

Get on their LinkedIn (https://www.linkedin.com/company/kokocares/) and comment on their posts! You may also want to tag the company's co-founders in your comments - their accounts are listed on the company page.

There's no way to reach support through chat, and commenting on a company's LinkedIn posts / tagging the people responsible is the best way to get a quick response to a sensitive issue - as their investors and research funders follow those posts, and companies take it seriously if safety issues are brought up in front of the people giving them millions of dollars.

To report it to the FCC for likely violating the COPPA law, regarding minors' safety and privacy online:

See Reblogs for further info & reporting instructions: Detailed description of COPPA law and Kokobot's presumed violations, plus detailed reporting instructions

But quick links: FCC reporting website and email hotline: [email protected]

Seriously, if you've taken the time to read this far, please please please take one more minute to file a report! It won't get addressed if all we do is reblog this, we need to get this in front of Tumblr Staff / The FCC / Koko's investors to get this meaningfully addressed.

Blocking and reporting the bot as spam isn't enough IMO - people have been doing that for years from the looks of the tag

----------

Reccomended reading in reblogs:

dropattackbear's discovery of what Koko is using the harvested data for (Machine Learning training data for automated content moderation services)

winderlylandchime (a licenced clinical psychologist's) explanation of privacy / ethics considerations around mental health services

thatsmimi's post on the dangers of letting minors act as a suicide / self-harm resource

My additions on their investors, leadership board, and their current job opening

----------------------------------------------------

Legal Disclaimer since tech companies LOVE lawsuits:

The views, thoughts, and opinions expressed in the text belong solely to the author, and not necessarily to the author's employer, organization, committee or other group or individual. This text is for entertainment purposes only, and is not meant to be referenced for legal, business, or investment purposes. This text is based on publically available information. Sources may contain factual errors. The analysis provided in this text may contain factual errors, miscalculations, or misunderstandings.