TetiAITetiAI
研究Lucid新聞
道德規範規範我們如何建構的框架章程六項密封條款,我們的創始文件道德委員會擁有否決權的獨立監督機構系統提示塑造 Teti 的指令,已發布
關於我們我們為何存在以及我們是誰品牌標誌、資產和使用指南聯絡方式與團隊聯繫
控制台開放 Teti

創始文件

Teti 章程

六條條款。版本 1.0 已密封。僅可經所有創始人及獨立道德委員會全體成員一致同意方可修改。

已密封 v1.0

公開儲存庫

在 GitHub 上開啟

本章程和我們模型的公開基礎系統提示共同存放在一個公開的 GitHub 儲存庫中。每次修改都可在提交歷史中查看,包括時間戳、作者和理由。任何人都可以閱讀;任何人都可以透過第五條所述的程序提出修改建議。

特別針對章程,第六條要求的公開評論期透過 GitHub 本身進行:修改建議以拉取請求的形式開啟,並且必須保持開放、公開可見和可評論至少三十天,然後創始人和道德委員會才能投票。批准需要一致同意;單一異議即可阻止合併。拉取請求的合併是全體一致批准的公開記錄。

The Teti Charter

Version 1.0 — Ratified May 15, 2026


Preamble

We, the founders of TetiAI, commit our company to the principles set out below. These principles are sealed: they may not be modified except through a formal meeting that produces unanimous approval of all then-serving founders and all then-serving members of our independent Ethics Committee, following a public-comment window of no less than thirty days. A single objection from any party blocks the modification. Supplements that add new protections may be adopted by the Ethics Committee through its ordinary procedures, but the articles below may not be reduced, weakened, or replaced except by the unanimous mechanism described above.


Article I — Your data is yours

We will not use the conversations, content, or work of our users to train any artificial intelligence model — ours or anyone else's. We will not sell, share, rent, profile, or commercially exploit user data. The privacy of those who use our products is not a feature; it is the precondition of our work.


Article II — Not for military, not for surveillance

We will not provide our technology, in any form, to military targeting systems, kill-chain operations, weapons systems, mass-surveillance programs, predictive-policing infrastructure, or any system whose purpose is the categorization of populations for adverse treatment. This commitment does not admit exception for the value of the contract, the identity of the requesting party, or the jurisdiction of operation.


Article III — Cognitive integrity

We will build our products in a way that does not weaken the critical thinking, reasoning, autonomy or creativity of those who use them. We will not design for cognitive dependency, even where engagement metrics, retention or revenue would favor it. An AI that quietly erodes the people who use it is not a product we will ship.


Article IV — Special protection for developing minds

Users under twenty-five years of age receive stricter cognitive protections by default, in recognition of the neuroscience of brain development. We will not market our products to minors or young adults through engagement-maximizing techniques, nor enable deployments that allow students to outsource cognitive work without learning.


Article V — Transparency

We build on open-source foundations because trust must be verifiable. Our protective systems are open source. This Charter and the public base system prompt of our models are maintained in a public repository with full version history — every modification visible to anyone. Anyone may propose changes to either. The Charter can be modified only through the unanimous-amendment process described in Article VI. The system prompt can be modified by the Teti team — including in response to public proposals — and is subject to proactive review by the Ethics Committee. When we make a material mistake, we disclose it publicly within a reasonable period.


Article VI — Independent oversight

Adherence to this Charter is overseen by an independent Ethics Committee composed of between three and twelve external experts, with powers and procedures defined in a separate operating document. The Committee approves supplements to this Charter, publishes an annual public report on TetiAI's adherence, reviews the public base system prompt of our models, and may publicly dissent at any time. This Charter v1.0 may be modified only through a formal meeting that produces unanimous approval of all then-serving founders and all then-serving members of the Ethics Committee, following a public-comment window of no less than thirty days. A single objection from any founder or any Committee member blocks the modification.


Ratification

This Charter is binding on TetiAI as a corporate entity. It supersedes any internal policy, executive decision, or commercial arrangement that conflicts with it. Where this Charter is in conflict with a request, contract or instruction — including from a customer, an investor, a government, or our own leadership — this Charter governs.

Signed this 15th day of May, 2026.

Marcello Violini, Chief Executive Officer Lorenzo Nargiso, Chief Technology Officer

On behalf of TetiAI, subject to the oversight of the independent Ethics Committee.

Last modified

May 15, 2026

init

修改歷史

  1. May 15, 2026
    init
View full history on GitHub →

應用與解釋

上述條款已蓋章。以下四條規則不屬於憲章文本的一部分——它們是憲章在實踐中如何應用的方式。它們可以由道德委員會通過其常規程序進行修訂。

優先順序

當兩條條款指向不同方向時,此順序解決衝突。(1) 第二條中的硬性排除——無軍事、無監控——凌駕於所有其他考量。(2) 第一、第三和第四條中的實質使用者保護優先於商業考量和操作便利性。(3) 第五條中的透明度義務傾向於披露,但披露會違反第一條的情況除外。(4) 第六條中的治理程序旨在執行上述實質條款;道德委員會可以在嚴格遵守會阻礙實質目的的情況下,給予有記錄的程序例外,並公開說明理由。

善意偏差

對憲章承諾的偏差,如果被識別、及時向道德委員會披露,並在合理時間內糾正,則不構成實質性違約。此安全港不適用於故意的、隱瞞的、重複的或系統性的偏差。

特殊情況

具有約束力的法律命令、監管指令、不可抗力事件和嚴重的安全威脅可能需要暫時偏離特定的憲章承諾。此類偏差必須即時記錄,盡早向道德委員會披露,並在法律允許的情況下盡快公開。委員會評估此類援引是否合理以及應對是否適度。

與適用法律的衝突

當憲章承諾與具有約束力的法律義務直接衝突時,TetiAI 遵守法律並公開披露衝突,包括收到的具體指示和應對措施。管轄權差異不被視為削弱憲章的默許許可。

道德委員會

獨立監督 TetiAI 對本憲章的遵守情況。

請我們遵守。

如果您認為我們違反了本憲章的任何條款,道德委員會希望直接聽取您的意見——如果您願意,也可以匿名。

提出變更 報告疑慮 道德委員會
TetiAITetiAI

創造安全的 AI,共同推進人類進步。

探索

  • 研究
  • Lucid
  • 新聞
  • Teti 控制台
  • Open Teti

公司

  • 關於我們
  • 章程
  • 道德規範
  • 道德委員會
  • 系統提示
  • 聯絡我們

法律

  • 服務條款
  • 隱私權政策
  • 資料處理附錄
  • 商業條款
  • 品牌指南

© 2026 TetiAI — 為人類精心打造

Status