This morning I witnessed the first US criminal trial on existential risk from AI.
Surprisingly, the judge did NOT initially dismiss Wynd Kaufmyn's necessity defense arguments for blocking the doors of OpenAI, to prevent an existential threat. The judge wants to hear more (1/) 🧵
The prosecutor's main argument is that granting a necessity defense here sets a dangerous precedent.
Anyone could cite "some esoteric existential threat" to justify interfering with a company, eroding the private property and trespass rules society runs on. (2/)
This morning I witnessed the first US criminal trial on existential risk from AI.
Surprisingly, the judge did NOT initially dismiss Wynd Kaufmyn's necessity defense arguments for blocking the doors of OpenAI, to prevent an existential threat. The judge wants to hear more (1/) 🧵
This was just the first hour and a half of the trial, that is currently restarting at 1:30pm at the superior court of california, room 503.
This trial will be happening this week and could run until June 17.
Hearing a judge talk about existential risk from AI (instead of dismissing it completely like in Musk v. Altman) felt like the start of a new chapter of history.
I highly recommend people coming to watch the trial, which will involve high-profile expert testimonies, including potentially Max Tegmark (@tegmark) and Stuart Russell.
Case number: CRI-25404422 (6/6)
Another point of contention has been the nb of doors that the OpenAI building had at the time.
By law, depending on the square footage, OpenAI would have been required to have two doors for fire hazard reasons. So employees would then still have been able to come in and out (5/)
The prosecutor stressed a necessity defense is meant for extreme circumstances, and argued this case is "the exact opposite end of that structure"
But the judge didn't immediately dismiss necessity defense. He just said the defense will have to provide "substantial evidence"(3/)
The prosecutor's main argument is that granting a necessity defense here sets a dangerous precedent.
Anyone could cite "some esoteric existential threat" to justify interfering with a company, eroding the private property and trespass rules society runs on. (2/)
This morning I witnessed the first US criminal trial on existential risk from AI.
Surprisingly, the judge did NOT initially dismiss Wynd Kaufmyn's necessity defense arguments for blocking the doors of OpenAI, to prevent an existential threat. The judge wants to hear more (1/) 🧵
Users are excited that a judge allowed an AI existential risk defense in the OpenAI protest trial because it marks the start of a new chapter.