2025年8月31日，一宗震動全球人工智慧領域的商業機密糾紛正式曝光。馬斯克旗下的xAI公司向加州法院提起訴訟，指控前核心成員李學辰（Xuechen Li）在離職前盜取公司整個代碼庫及商業機密，隨後跳槽至競爭對手OpenAI。消息一出，馬斯克本人也在社交媒體上親自證實此事，並表示相關行為已經被操作日誌和李學辰本人承認。

整起事件的時間脈絡相當清晰。根據訴狀內容，2025年7月25日，李學辰在正式離職前三天，利用自己全技術棧的訪問權限，將大量涉及Grok模型開發的核心機密數據加密後上傳至個人系統，並同時刪除操作日誌、重新命名檔案，以試圖掩蓋行蹤。7月28日，他正式辭職離開xAI，此前已經套現近700萬美元的公司股權。到8月11日，xAI在安全審查中發現異常，要求李學辰配合調查，但他卻修改關鍵賬號密碼以阻撓。直到8月14日，他在律師陪同下承認了盜取資料的事實，但仍刻意隱瞞部分文件。僅僅五天後，即8月19日，他便加入OpenAI，被指涉嫌攜帶的技術可能為對手節省「數十億美元的研發成本」。

在法律層面，xAI向法院提出四項指控，包括違反保密協議、侵犯商業秘密、違反加州電腦數據相關法規，以及欺詐行為。訴訟要求李學辰進行賠償，禁止進入競爭對手公司，並交出所有涉案設備與帳號。這一訴求不僅針對個人責任，也意在防止核心技術外流進一步損害公司利益。

李學辰的背景使這起事件更具爭議性。他是xAI創始團隊的早期二十名工程師之一，擁有斯坦福博士學位，曾深度參與 Grok 模型的研發，並掌握公司內部幾乎所有技術資料。正因如此，他能夠運用權限優勢進行資料竊取，採取加密傳輸、清理日誌、簽署虛假聲明等手段來掩蓋行為。這也暴露出xAI在核心人員安全管理與機密保護上的漏洞。

此案的曝光，不僅凸顯 AI 行業在商業秘密保護上的脆弱性，也讓外界再次關注「開源與閉源」的戰略爭論。隨著人才競爭日趨激烈，如何平衡技術開放與專利保護，成為產業內的熱點話題。雖然 OpenAI 尚未被正式列為被告，但社會輿論普遍質疑其在此次「挖角」事件中的角色，甚至戲謔性地發問：「為什麼總是 OpenAI？」

截至2025年9月1日，李學辰與OpenAI對外仍保持沉默，案件則在美國加州法院持續審理中。這場風波不僅關乎一名工程師的個人責任，更可能演變為兩大AI巨頭之間的正面衝突，牽動著全球人工智慧產業的未來走向。

On August 31, 2025, a case that shook the global artificial intelligence industry came to light. Elon Musk’s company xAI filed a lawsuit in California against former core member Xuechen Li, accusing him of stealing the company’s entire codebase and trade secrets before leaving to join competitor OpenAI. Musk personally confirmed the matter on social media, stating that the misconduct had been corroborated by system logs and Li’s own admission.

The timeline of events is clear. According to the lawsuit, on July 25, 2025, just three days before his resignation, Li used his full-stack access privileges to encrypt and upload vast amounts of xAI’s confidential data—including material tied to the development of the Grok model—to his personal systems. He then deleted activity logs and renamed files in an effort to conceal his actions. On July 28, he formally resigned from xAI, having already cashed out nearly $7 million worth of company equity. On August 11, xAI’s security review detected irregularities and requested his cooperation, but Li instead changed critical account passwords to obstruct the investigation. By August 14, accompanied by his lawyer, Li admitted to stealing company data, though he withheld details about certain stolen files. Merely five days later, on August 19, he joined OpenAI, with the stolen technology allegedly capable of saving the competitor “billions of dollars in R&D costs.”

Legally, xAI has filed four charges: breach of non-disclosure agreement, misappropriation of trade secrets, violation of California computer data laws, and fraud. The company is seeking damages, an injunction barring Li from working with competitors, and the surrender of all devices and accounts tied to the incident. The suit aims not only to hold Li personally accountable but also to prevent further leakage of proprietary technology.

Li’s background adds another layer of controversy. As one of xAI’s earliest engineers—among its founding team of about 20—he held a Ph.D. from Stanford and played a central role in developing the Grok model. With unrestricted access to the company’s internal systems, he was able to exploit his privileges, encrypt data transfers, scrub logs, and even sign false statements to cover his tracks. This has exposed serious gaps in xAI’s internal safeguards for managing key personnel and protecting trade secrets.

The case has raised broader concerns beyond the courtroom. It highlights the fragility of trade secret protection in the AI sector and reignites debate over the balance between open-source and closed-source strategies. With competition for talent intensifying, the incident underscores the risks companies face when top engineers move between rivals. Although OpenAI has not been named a defendant, public speculation has mounted regarding its role in the “poaching” incident, with online commentary wryly asking: “Why is it always OpenAI?”

As of September 1, 2025, both Li and OpenAI have remained silent, while the case proceeds in California court. The dispute is more than a matter of individual misconduct—it could escalate into a direct confrontation between two of the AI industry’s most influential players, with significant implications for the future direction of the global AI race.