Not All Anquan Is the Same: A Terminological Proposal for Chinese Computer Science and Engineering

Technical Summary: "Not All Anquan Is the Same: A Terminological Proposal for Chinese Computer Science and Engineering"

Problem Statement
In the Chinese computer science and engineering (CSE) community, the English terms safety and security have historically been translated using the same character: 安全 (anquan). While this convention offers brevity in casual discourse, the paper argues it creates a persistent "conceptual compression" that hinders rigorous scientific communication, standard interpretation, and risk analysis.

The core issue is that safety and security represent distinct engineering logics with different risk models, evidence requirements, and organizational responsibilities:

• Safety (safety) concerns a system's tendency to cause unintentional harm to its environment (people, property, mission) due to failures, performance limitations, or foreseeable misuse. It asks: "Will the system harm the environment?"
• Security (security) concerns a system's vulnerability to intentional, adversarial threats from its environment (attacks, misuse, unauthorized access). It asks: "Will the environment harm the system?"

The paper posits that collapsing these distinct concepts into a single Chinese term obscures critical distinctions in fields like functional safety (ISO 26262), expected functional safety (SOTIF), cybersecurity, and AI assurance. This ambiguity complicates the formulation of "assurance cases" (structured arguments for trustworthiness), as a single term cannot simultaneously support arguments about hazard analysis (e.g., FMEA) and threat modeling (e.g., attack trees) without relying on context-dependent interpretation, which fails in formal documentation and automated indexing.

Methodology
The paper employs a conceptual and terminological analysis rather than empirical experimentation. Its methodology involves:

1. Conceptual Mapping: Analyzing international and domestic standards (e.g., ISO/IEC 27000, IEC 61508, ISO 26262, NIST CSF) to delineate the distinct problem structures, risk sources, and evidence types associated with safety versus security.
2. Linguistic Analysis: Evaluating the cognitive impact of terminology on research communities, drawing on the work of Bowker and Star regarding classification systems and Wittgenstein's views on language limits.
3. Case Study of AI Assurance: Examining recent developments in AI assurance (e.g., Assurance 2.0, security-informed safety) to demonstrate how the lack of distinct terminology impedes the construction of "challengeable arguments" where specific defeat conditions (e.g., a supply chain attack defeating a safety claim) can be precisely identified.
4. Terminological Proposal: Proposing a specific translation strategy that respects existing legal and standard nomenclature while introducing a new distinction for academic and engineering discourse.

Key Contributions
The paper's primary contribution is a structured proposal for a "dual-track" terminological practice in Chinese CSE:

1. Terminological Distinction:

   • Safety should continue to be translated as 安全 (anquan) or 安全性 (anquan-xing).
   • Security should be translated as 安保 (anbao) or 安保性 (anbao-xing).
   • The term 安保 is selected because it already carries connotations of protection, guarding, and adversarial response in public governance and international relations (e.g., "security liaison mechanisms"), distinguishing it from the broader, often passive connotation of 安全.

2. Dual-Track Implementation Strategy:

   • Official Contexts: The paper explicitly acknowledges that existing laws (e.g., Cybersecurity Law) and national standards (e.g., GB/T 22239) use 安全 for security. It does not propose rewriting these.
   • Academic/Engineering Contexts: In new papers, textbooks, and technical documents, authors should use 安保 for security concepts. When citing official titles, authors should use the official name but clarify the conceptual distinction in the text (e.g., "Information Security (officially Information Anquan; this paper refers to it as Information Anbao for conceptual clarity)").

3. Conceptual Framework for Interrelation:
   The paper argues that clear terminology is a prerequisite for effective safety-security co-assurance. It cites frameworks like security-informed safety and Safety-Security Assurance Framework (SSAF), noting that these approaches require distinct concepts to manage their interfaces. If the terms are merged, the specific mechanisms by which a security failure defeats a safety argument (or vice versa) become linguistically invisible.

4. Proposed Glossary:
   The paper provides a suggested glossary (Table 3 in the text) mapping key terms (e.g., functional safety, SOTIF, AI safety, AI security, security case, safety case) to the proposed Chinese terminology to facilitate immediate adoption.

Results and Claims
The paper does not present experimental results or performance metrics. Instead, its "results" are logical and structural:

• It demonstrates that the current translation overload creates a "cognitive anchor" problem where complex concepts like AI assurance and safety-security co-assurance collapse into ambiguous phrases like "AI safety" or "safety argument," losing the ability to distinguish between hazard-based and threat-based risks.
• It establishes that the distinction between "system harming environment" (safety) and "environment harming system" (security) is fundamental to modern cyber-physical systems (e.g., autonomous vehicles, industrial control) and cannot be adequately managed with a single term.
• It claims that adopting 安保 for security will reduce communication costs, improve the precision of risk registers and requirement documents, and enable more rigorous scientific argumentation in assurance cases.

Significance
The paper frames its significance as a necessary step in "terminological governance" for the Chinese CSE community. It argues that language shapes the "limits of the world" (Wittgenstein) and that without a stable term for security distinct from safety, the community struggles to index, teach, review, and standardize these critical fields separately.

The significance is modest but practical: it does not demand an immediate overhaul of national standards but proposes a "low-friction" adoption path for the academic and engineering community. By creating a stable, searchable, and distinct term (安保性), the paper aims to enable researchers to formulate questions, build evidence, and review arguments with the necessary precision required for the next generation of software-intensive, AI-driven, and safety-critical systems. The ultimate goal is to ensure that "not all anquan is the same" is a recognized and actionable principle in Chinese technical discourse.