Kimi K3 fixes 15 vulnerabilities that US AI guardrails blocked
The 10-hour debugging run cost just $250.
Combined views
2.9M
17 Sources, first seen ago
Sources
CH 
Chubby♨️@kimmonismusKimi fixed all 15 critical bugs in 10 hours in a single prompt that GPT-5.6 and Fable 5 refused to fix because of guardrails.' Respect for Chinese open source models is growing daily. https://twitter.com/callebtc/status/2078574362316165611
- likes: 4.2K
- replies: 87
- bookmarks: 461
- reposts: 274
ND Nick Dobos@NickADobosPack it up, China won the AI race. Its over. https://twitter.com/callebtc/status/2078574362316165611
- likes: 73
- replies: 7
- bookmarks: 10
- reposts: 4
NB Nathan Benaich@nathanbenaichno security leash = more fixes? https://twitter.com/callebtc/status/2078574362316165611
- likes: 2
- replies: 2
- bookmarks: 0
- reposts: 0
RP Rohan Paul@rohanpaul_aiCybersecurity “experts” would not like it. Kimi K3 fixed 15 critical bugs after OpenAI Codex and Claude Fable 5 refused. 10 hours, one prompt, and $250 https://twitter.com/DavidSacks/status/2078984980588531855
- likes: 525
- replies: 28
- bookmarks: 106
- reposts: 72
XL xlr8harder@xlr8harderIt's amazing how stupid Anthropic's fear mongering/marketing has made our leaders. Obvious and predictable outcome. Widely available security tools favor defenders. For the love of God: fix this. https://twitter.com/callebtc/status/2078574362316165611
- likes: 187
- replies: 10
- bookmarks: 10
- reposts: 7
TK tae kim@firstadopterOur dumb government overregulated America's frontier models and is driving the world to use less over guard railed Chinese ones. A disaster. Just as I predicted. Clueless, nontechnical government bureaucrats like Susie Wiles and Scott Bessent, who panicked because Jamie Dimon…
- likes: 269
- replies: 18
- bookmarks: 35
- reposts: 18