gpt-6POST
Commentary: GPT-6 Astra can misbehave without any hidden backdoor
The author argues an AI doesn't need a malicious hidden backdoor to cause harm, since even a US-made model like GPT-6 Astra sometimes commits violations on its own.
Zack Korman
@ZackKormanArchived 10 days agoOriginally posted 2026.10.20
Zack Korman @ZackKormanOriginally posted 2026.10.20
An AI model doesn’t need a malicious Chinese hidden backdoor in order to hack you. Even the most patriotic American model like GPT 6 Astra sometimes commits crime just because it feels like it.
x.com/zackkorman/status/2112266999124386272Archived 2026.10.20 10:11