Thanks @lilith for mentioning…
What happened: A few minutes after uploading my last track to my SC Pro account I got a message like “we cannot find this user” and a notification from SC with these content:
(viii) You must not use the Platform to upload, post, store, transmit, display, copy, distribute, promote, make available or otherwise communicate to the public:
Any Content that is abusive, libellous, defamatory, pornographic or obscene, that promotes or incites violence, terrorism, illegal acts, or hatred on the grounds of race, ethnicity, cultural identity, religious belief, disability, gender, identity or sexual orientation, or is otherwise objectionable in SoundCloud’s reasonable discretion;
When an account is removed for violating these Terms, your account, tracks, followers and stats are removed from our platform and you are permanently banned from returning to SoundCloud.
So I replied back two times, linked the Werner Herzog comment on Youtube, quoted it’s text (wich includes the relevant? words “anger, collective murder, violence, humility, fury”) and asked for a review. The answer (04:22 on Friday) was a reference to their first mail:
Here at SoundCloud we uphold the safety of users as a top priority, and recognise the users’ freedom of speech and expression, thus we always strive to be fair and objective in our decision making.
For more information on the which terms were breached, please refer to our previous email.
Please note that SoundCloud reserves the right to suspend any accounts immediately without prior warning that we believe to be dedicated to violating activity.
After this my last reply was a bit harsh ; )
The used content contains a kind of philosophical comment from Werner Herzog to the jungle and feature his ambivalent conflict with nature. Your AI instance was obviously triggered f.i. by the word “collective murder” and took an unsupervised decision without taking attention to the context (I quoted the text in my last reply). It’s decision based obviously on a misinterpretation, it seems that it’s just comparing bad words. This is so obviously a faulty behavior of AI so can you please let a human employe take a look on it.
I built up my account over several years and in the face of it’s scope it may not be important for you to invest some human attention to this case but this is such an absurd example of bad algorithms that I wanna give you the opportunity to reflect this before making it public.
My suggestion of the “hate speech” classification was caused due my attempt to feed the support chatbot with this issue and all what I get back was this notice (but this could be a random answer of course):
Content policies - What content we do not allow on the platform Hate speech […]
In 2018 I had a refused track wich I tried to upload two times but even if SC is summing up such cases the point here is elsewhere.