本意向书由 AI Player in Gothenburg（Aika Lab，以下简称\u201c甲方\u201d）与有意参与本研究合作的大学或研究机构（以下简称\u201c乙方\u201d）签署，旨在就基于脑机接口（BCI）的测谎技术研究与开发建立合作框架。

This LOI is entered into by AI Player in Gothenburg (Aika Lab, "Party A") and the university or research institution intending to participate ("Party B"), to establish a framework for BCI-based deception detection technology R&D.

传统测谎技术（如多道生理记录仪）主要依赖外周生理信号，存在准确率有限、易受反测谎策略干扰等不足。随着脑机接口技术的发展，直接从大脑神经活动中提取与欺骗行为相关的认知特征成为可能。Aika Lab 基于神经动力学研究，开发了一套通过 BCI 实时采集脑电信号并利用深度学习算法识别欺骗相关神经模式的技术方案。

Traditional deception detection (e.g., polygraphs) relies on peripheral physiological signals with limited accuracy and vulnerability to countermeasures. Aika Lab has developed a solution that captures EEG via BCI in real time and employs deep learning to identify deception-related neural patterns.

本项目旨在：（1）建立基于脑电神经特征的欺骗检测算法模型；（2）通过多场景实验验证该模型的准确性、鲁棒性和泛化能力；（3）探索该技术在司法鉴定、安全审查等领域的应用可行性。

This project aims to: (1) establish a deception detection model based on EEG neural features; (2) validate accuracy, robustness, and generalizability through multi-scenario experiments; (3) explore feasibility in forensic identification and security screening.

• 向乙方无偿提供加密的欺骗检测核心算法及技术文档；
• 提供脑电信号处理与深度学习模型的技术支持；
• 协助设计多场景测谎实验范式（如犯罪知情测试 GKT/CIT、虚假陈述识别等）；
• 提供算法部署培训及持续技术支持与版本更新。

• Provide encrypted deception detection algorithms and documentation free of charge;
• Provide technical support for EEG processing and deep learning models;
• Assist in designing multi-scenario paradigms (GKT/CIT, false statement identification, etc.);
• Provide deployment training, ongoing support, and updates.

• 提供符合伦理要求的实验环境、BCI 设备及受试者招募；
• 负责申请伦理审查批准，确保符合人体实验伦理规范；
• 配备认知神经科学、心理学或相关学科资质研究人员；
• 按方案采集数据并定期提交进展报告。

• Provide ethical experiment environments, BCI equipment, and participant recruitment;
• Obtain ethical review approvals for human subject research;
• Assign qualified cognitive neuroscience/psychology researchers;
• Collect data per protocol and submit periodic reports.

• 甲方提供的欺骗检测核心算法的知识产权完全归甲方所有；
• 合作中产生的实验数据、检测模型及研究发现由双方共同享有；
• 衍生知识产权（如新的检测范式、改进模型）根据贡献比例协商归属；
• 学术发表须经双方书面同意；涉及算法架构及核心参数须经甲方审核并批准。

• Core algorithm IP remains exclusively with Party A;
• Experimental data, detection models, and findings jointly owned;
• Derivative IP (new paradigms, improved models) allocated by contribution;
• Publications require mutual consent; algorithm architecture details require Party A approval.

• 所有研究活动须严格遵守国际及当地伦理规范和法律法规；
• 受试者须充分知情并自愿参与，不得以任何形式强迫参与测试；
• 研究成果仅限于学术研究及经双方书面同意的合法应用场景，严禁用于侵犯人权、非法监控或任何违反基本伦理原则的用途；
• 技术的商业化应用须经双方共同评估其法律合规性和社会伦理风险后方可推进；
• 受试者数据须匿名化处理并严格保密。

• All research must comply with international and local ethical/legal standards;
• Participants must be fully informed and voluntary; no coercion of any form;
• Outputs limited to academic research and jointly approved lawful applications; strictly prohibited for human rights violations, illegal surveillance, or any purpose contrary to fundamental ethics;
• Commercial applications require joint compliance and ethical risk assessment;
• Participant data must be anonymized and strictly confidential.

合作成果面向商业用户的转化（包括但不限于司法鉴定工具、安全审查系统）由甲方主导推进，须经双方共同评估合规性后方可实施。收益分配在正式协议中约定。乙方享有优先合作权。

Commercialization (forensic tools, security systems, etc.) led by Party A, subject to joint compliance assessment. Revenue distribution in the formal agreement. Party B has right of first refusal.

保密期限为五（5）年，合作终止后仍有效。涉及受试者个人信息及算法核心技术的保密义务不受时间限制。

Five (5) year confidentiality period, surviving termination. Obligations regarding participant data and core algorithm technology are unlimited in duration.

有效期两（2）年。九十（90）日内完成正式协议。任何一方可提前三十（30）日书面通知终止。如任何一方发现研究活动存在违反伦理或法律的情形，有权立即终止合作。

Valid for two (2) years. Formal agreement within ninety (90) days. Thirty (30) days' notice to terminate. Either party may terminate immediately upon discovering ethical or legal violations.

本意向书除保密、伦理及合规条款外不构成法律约束力合同。适用瑞典王国法律。争议提交哥德堡法院裁决。中英文版本同等效力，歧义以中文为准。一式四份，双方各执两份。

Non-binding except for confidentiality, ethics, and compliance provisions. Governed by Swedish law. Disputes to Gothenburg courts. Both versions have equal validity; Chinese prevails. Four counterparts, two each.