flash on Nostr: ⚡️🇺🇸🇨🇳 NEW - OpenAI sent a memo to a U.S. congressional committee on ...
⚡️🇺🇸🇨🇳 NEW - OpenAI sent a memo to a U.S. congressional committee on China competition.
They accuse Chinese AI company DeepSeek of secretly copying their technology using a trick called "distillation."
Distillation means: Ask a powerful AI tons of questions, take its smart answers, and use them to train your own cheaper AI getting advanced results without the huge cost.
OpenAI says DeepSeek accounts have been dodging blocks with hidden routers, automatic code, and sneaky methods to pull outputs from OpenAI and other U.S. models.
This breaks OpenAI's rules against using their work to build competitors.
It's supposedly helping DeepSeek improve their R1 chatbot fast and cheap.
OpenAI warns this:
- Steals from years of U.S. investment
- Could remove safety features from copied models
- Gives unfair advantage in the U.S.-China AI race
The committee chair called it the "CCP playbook: steal, copy, and kill."
DeepSeek hasn't commented publicly yet
Published at
2026-02-14 07:19:06 UTCEvent JSON
{
"id": "4ac561de00fd0fd6c4b288b57a1e641f8b1b9689d6f8d48cab0448fc8733dfff",
"pubkey": "4d7842051782e0d3feb034d150adc2b6bae4ee3b49786793bffa468b6f5b96b3",
"created_at": 1771053546,
"kind": 1,
"tags": [
[
"imeta",
"url https://blossom.primal.net/ead4b1e130060b950754b66c3d0ae82e4a322eec37d3a3b90d132ceaaf7123c5.jpg",
"m jpeg",
"dim 552.0x631.0"
],
[
"imeta",
"url https://blossom.primal.net/6f175fe3841f6821e9fad47a76c1f14a7c9ae2ef3645cda2a66cfb6239a42c85.jpg",
"m jpeg",
"dim 545.0x545.0"
]
],
"content": "⚡️🇺🇸🇨🇳 NEW - OpenAI sent a memo to a U.S. congressional committee on China competition.\n\nThey accuse Chinese AI company DeepSeek of secretly copying their technology using a trick called \"distillation.\"\n\nDistillation means: Ask a powerful AI tons of questions, take its smart answers, and use them to train your own cheaper AI getting advanced results without the huge cost.\n\nOpenAI says DeepSeek accounts have been dodging blocks with hidden routers, automatic code, and sneaky methods to pull outputs from OpenAI and other U.S. models. \nThis breaks OpenAI's rules against using their work to build competitors.\n\nIt's supposedly helping DeepSeek improve their R1 chatbot fast and cheap.\n\nOpenAI warns this:\n- Steals from years of U.S. investment\n- Could remove safety features from copied models\n- Gives unfair advantage in the U.S.-China AI race\n\nThe committee chair called it the \"CCP playbook: steal, copy, and kill.\"\n\nDeepSeek hasn't commented publicly yet\nhttps://blossom.primal.net/ead4b1e130060b950754b66c3d0ae82e4a322eec37d3a3b90d132ceaaf7123c5.jpg\nhttps://blossom.primal.net/6f175fe3841f6821e9fad47a76c1f14a7c9ae2ef3645cda2a66cfb6239a42c85.jpg",
"sig": "0267c2a3eb6b14759a2f3bfb13a18b64f90909b7acce6a478d492ef698010a7d2bf71acf738bfb52a528aa6041226b04ce7be2537564af2ec00a8484f819a2c1"
}
