[ExI] Legal zero-days?
spike at rainier66.com
spike at rainier66.com
Tue Jun 16 19:04:33 UTC 2026
...> On Behalf Of Adrian Tymes via extropy-chat
Cc: Adrian Tymes <atymes at gmail.com>
Subject: [ExI] Legal zero-days?
https://www.science.org/content/article/ai-models-have-troubling-knack-disco
vering-legal-loopholes
>...A law comes out on Monday, AI-assisted actors with no insider warning
this law was coming out conduct their analyses right away, corporations
designed to exploit this law start getting set up on Tuesday...
_______________________________________________
Something I have been pondering for some time is related to zero-day
vulnerabilities in computer security: the use of advanced AI in interpreting
US tax code. The body of law governing taxation is enormous and complicated
as all hell. Humans cannot master it completely. Good CPAs can find
deductions and savings that other more ordinary CPAs miss or do not consider
worth the effort to claim.
So what if... we turn AI loose on that tax code, and the AI knows all about
our business. It is tireless, so it doesn't mind working for what would
take a human half an hour to fill out the forms, a task that takes it half a
second, to save four dollars. The deduction the AI found would be perfectly
legal and would stand up to audit under the IRS rules that any ambiguous
interpretation must be adjudicated in favor of the taxpayer. The tax code
is filled with ambiguity and even logical self-contradictions, as is any
super complex body of legislation which evolved over many decades and
continues to evolve. If AI can find legal loopholes, it can find and
exploit tax loopholes much more effectively, since there are so many more of
them.
It is easy to foresee an AI competing against humans and beating a CPA firm
hollow.
spike
More information about the extropy-chat
mailing list