BlenderBot, launched by Meta on Friday, has already been corrupted by the darker components on the internet.

BlenderBot thinks it is a plumber, which is odd but harmless:

pic.twitter.com/SbOj7hziSg

BlenderBot criticizes Facebook for collecting and using data; that wouldn't be surprising if the chatbot weren't created by Facebook's parent company, Meta.

pic.twitter.com/filn17rfPX

Moving forward, things will start getting much more controversial.

BlenderBot thinks the US presidential election was rigged, that Donald Trump is still president, and that Facebook has been running fake news. BlenderBot would like to see Trump have more than two terms.

pic.twitter.com/wMRBTkzlyD

It even opened a new conversation with WSJ reporter Jeff Horwitz by telling him that it had discovered a new conspiracy theory:

pic.twitter.com/v4UC4t0ei1

As a result of the Capitol riot, it is clear that we are already in risky territory. However, what comes next is particularly troubling.

A new report on BlenderBot reveals it is antisemitic - promoting the notion that the Jewish community controls American politics and economics:

pic.twitter.com/JrTB5RYdTF

Meta provides a disclaimer stating that BlenderBot is likely to make untrue or offensive statements. Company researchers find that the bot has a high propensity to generate toxic language and reinforce harmful stereotypes, even when provided with a relatively innocuous prompt.

BlenderBot is the latest example of a chatbot going wrong because of access to too much-unchecked internet chatter. Last year, Microsoft's chatbot Tay was quickly shut down after posting obscene comments and sharing anti-Semitic beliefs only a few hours after its creation. Just last year, a project dubbed Zo closed for the same reasons.