It causes the end of humanity---AI corporate control would generate insurmountable wealth that allows it to control society---it’s existential

Roman V. Yampolskiy, "Human Indignity: From Legal AI Personhood to Selfish Memes," arXiv:1810.02724, 10-2-2018, https://arxiv.org/abs/1810.02724

5. Human Safety

A lot has been published on different risks associated with advancement of artificial intelligence [36-43], but less specifically on dangers from AI controlled corporate entities. Nothing in our current laws would prevent formation of a malevolent corporation (or corporate virus) with memetic payload of subjugating or exterminating humanity through legal means and political influence. In addition to legal enslavement of people via below living-wage salary, such corporations could support legal change in minimum wage and pension laws as well as provide opposition to wealth redistribution and Universal Basic Income/Universal Basic Assets [44, 45]. This is particularly easy to accomplish because of Supreme Court decision in Citizens United VS FEC [46], permitting unrestricted donations from corporations to politicians under the guise of free speech, making it possible to convert financial wealth to political power.

This leads us to recognize an additional existential risk (X-risk) [47], from extreme wealth. Wealth inequality is already recognized as a problem for democratic institutions [48], but super-rich corporate AIs (dollar trillionaires) would take that problem to the next level. They could accumulate such unprecedented levels of wealth via unfair business practices such as predatory pricing, having access to free physical and cognitive labor from direct control of automation, permitting them to undermine completion and achieve monopoly status in multiple domains and other multiple resources. Additionally, such entities could engage in super-long term investment getting compound interest for hundreds of years. For example, a million dollars invested for 150 years at the same rate of return as observed over the last hundred years would grow to 1.6 trillion inflation-unadjusted dollars, creating super-rich artificial entities.

If EAs become intellectually indistinguishable from people, meaning could pass an unrestricted Turing Test [49] their capacity to self-replicate could be used to drain resources from legitimate corporations, for example via click-fraud [50, 51] from Google. Also, they will be able to create their own super successful companies with alternative populations comprised of billions of EA users indistinguishable from real people but paid for by advertisers as if they were genuine clients.