Expired50% confidenceFactExact time
According to OpenAI's Preparedness Framework, Soul is close to but has not crossed the critical red line in the cybersecurity dimension.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/8/2026
First Seen
Valid until: 10/6/2026(expired)
Sources
GPT-5.6三版本解析：OpenAI首次让政府审核AI模型
bilibili五里墩茶社6/28/2026
Related Claims
Unverified黄仁勋认为开源模型的透明性（代码和权重公开可供全球研究者审查）本身就是安全的最佳保障63% similarUnverifiedMatt Green推断OpenAI与Anthropic极可能都使用单一全局密钥加密所有客户端推理数据，属于密钥管理反模式60% similarUnverifiedCybersecurity capability can be seen as a byproduct of strong coding ability; models that can understand complex code logic and locate bugs inherently have the potential to discover security vulnerabilities.58% similarUnverifiedIn the announcement of Codex's remote control feature, OpenAI specifically emphasized the word 'securely,' indicating targeted design in areas such as permission management, communication encryption, and operation auditing.58% similarUnverifiedTraditional cybersecurity measures such as firewalls, encryption, and access controls are nearly impossible to directly repurpose when facing this new type of threat known as the cognitive attack surface.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/236067API
curl https://kongchang.com/api/v1/knowledge/claims/236067MCP
get_claim(id=236067)