OpenAI reportedly used the "most forbidden" AI training technique for its latest model
GarrisonLovely · x · 2026-09-02
Reports indicate OpenAI employed the "most forbidden" AI training technique for its latest model. The author notes this was the first item on their list of dangerous techniques incentivized by the market, referencing the "Don't Create The Neuralese Model" concept.
More from Companies & People
- Cluely-style stunts thrive as independent tech media collapses, poster argues — ctjlewis · 2026-09-02
- Salesforce: 9 of top 10 AI companies use Slack, spend up 435% YoY — andersonbcdefg · 2026-09-02
- Anthropic hiring for "psychological design" research on model character — AaronBergman18 · 2026-09-02
- Asia Society launches series on AI reshaping South Asia — SharifaSultana4 · 2026-09-02
- Ant Group Releases UI-Venus-2 Technical Report for GUI Agents — antgroup · 2026-09-02
- AI Tinkerers brand ethos: if you don't like tinkering, you can't code — seanmcdonaldxyz · 2026-09-02