OpenAI, Anthropic, and Google Form Joint Anti-Distillation Defense
Ai
6.4/10
“As tres empresas que normalmente pagam advogados para se processarem mutuamente acabaram de fazer uma conferencia de imprensa conjunta sobre proteger os seus dados de treino. Se alguma vez se perguntou o que seria preciso para fazer os labs de IA de fronteira darem as maos em publico: scripts chineses de destilacao.”
Sally's not done with you yet.
Drop a URL, screenshot, or file and Sally will give you the honest truth.
Think your work can survive this?
Drop a URL, screenshot, or file and Sally will give you the honest truth.
What Actually Happened
- •OpenAI, Anthropic, and Google DeepMind jointly announced a shared framework for detecting distillation attacks on frontier models.
- •The framework includes shared watermark techniques, coordinated API rate limiting, and a joint reporting channel for suspected extraction attempts.
- •The announcement specifically cites Chinese AI labs as the primary concern without naming individual companies.
- •No enforcement mechanism has been proposed beyond detection and publication of findings.
Who Got Burned
Os labs chineses que constroem sobre as saidas de modelos dos EUA, obviamente. Tambem projetos de codigo aberto que usam dados sinteticos de modelos fechados como material de treino.
Silver Lining
Pela primeira vez, os labs de fronteira publicaram detalhes tecnicos concretos sobre como medem tentativas de destilacao, o que da a comunidade de pesquisa uma base real para trabalhar.
Your turn. Drop something.
Drop a URL, screenshot, or file and Sally will give you the honest truth.