Unverified60% confidenceFactExact time
In trade secret determinations in the AI field, model architecture may be protected by copyright, while training methods, data recipes, hyperparameter tuning strategies, and proprietary evaluation benchmarks are typically protected as trade secrets.
2
Sources
60%
Confidence
Long-term
Relevance
7/11/2026
First Seen
Sources
Apple Sues OpenAI Over Trade Secret Theft: The Deep Logic Behind a Broken Coopetition
hackernewshackernews7/10/2026
Related Claims
UnverifiedLegal AI products typically need to offer private deployment options, commitments against using data for training, and SOC 2 compliance certification in order to earn procurement trust from large law firms or enterprise legal departments.78% similarUnverifiedModel weights are the most core intellectual property of AI companies; leakage could be used to bypass safety alignment mechanisms or replicate proprietary capabilities.77% similarUnverifiedDevelopers should be cautious about publicly sharing projects that use AI to replicate commercial tools, pay attention to open-source alternatives, and understand relevant trademark law, copyright law, and unfair competition regulations.75% similarUnverified商业秘密保护、知识产权边界划定与人才流动规范正成为决定AI行业格局的关键变量74% similarUnverifiedThe EU AI Act requires AI developers to disclose summaries of copyrighted content included in training data.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/487965API
curl https://kongchang.com/api/v1/knowledge/claims/487965MCP
get_claim(id=487965)