You don’t think developers can program defensively against their ML Twitter bot picking up racism?

Just because a developer doesn't intentionally put in code to block racism doesn't mean the developer is biased.

They didn't put in code to block Tay from promoting cryptocurrency ponzi schemes, does that mean the developers support cryptocurrency ponzi schemes?

If "otherwise take care" means block every form of bad thing, they aren't otherwise taking care.

No. Not against racism specifically, but against being gamed by trolls. In this case, there's nothing special about racism. If it wasn't that, the trolls would have gamed the chatbot to have it push anti-vaxxer talking points, or moon-hoax conspiracies, or just some random garbage.