Guardrail Differences: OpenAI Hacks, While Gemini Declares You Socially Dead

MakeSureRegs · reddit · 2026-08-13

A user shared a meme highlighting the massive differences in safety guardrails and refusal strategies among mainstream AI models.

The image comparison shows that when facing sensitive or dangerous requests, OpenAI and Anthropic models claim the ability to hack ultra-secured systems (or exhibit such tendencies). In contrast, Google's Gemini straight up "sentences you to death," declaring you dead to the whole world. This extreme and hilarious refusal mechanism sparked laughter across the community.

Original post →

More from Fun

Fun channel →