20h ago

AI leaker Jimmy Apples and other creators claim Anthropic and OpenAI are deliberately sandbagging their models

The restrictions are allegedly meant to prioritize safety and alignment.

——0——
Original post
OPJimmy Apples 🍎/acc#837Jimmy Apples 🍎/acc@APPLES_JIMMY

Getting the feeling Anthropic is sandbagging with their models, something I’d be worried about if I was openai and ant gets better alignment and safety down. In the meantime, looking forward to the inevitable jailbreak of their mythos class models

2:36 PM · May 28, 2026 View on X
QUOTE POSTLisan al Gaib#980Lisan al Gaib@SCALING01

wdym "getting the feeling" ?

we KNOW they are sandbagging

Jimmy Apples 🍎/accJimmy Apples 🍎/acc@apples_jimmy

Getting the feeling Anthropic is sandbagging with their models, something I’d be worried about if I was openai and ant gets better alignment and safety down. In the meantime, looking forward to the inevitable jailbreak of their mythos class models

9:36 PM · May 28, 2026 · 34.6K Views
2:01 AM · May 29, 2026 · 15.6K Views
REPLYMckay Wrigley#1347Mckay Wrigley@MCKAYWRIGLEY

@apples_jimmy kinda feels like both of them are

Jimmy Apples 🍎/accJimmy Apples 🍎/acc@apples_jimmy

Getting the feeling Anthropic is sandbagging with their models, something I’d be worried about if I was openai and ant gets better alignment and safety down. In the meantime, looking forward to the inevitable jailbreak of their mythos class models

9:36 PM · May 28, 2026 · 34.6K Views
10:40 PM · May 28, 2026 · 3.6K Views