齊寶鑫：淺談當前形勢下的知識產(chǎn)權戰略面臨的主要挑戰和趨勢

#本文僅代表作者觀(guān)點(diǎn)，不代表IPRdaily立場(chǎng)#

“面對高人員流動(dòng)性帶來(lái)的商業(yè)秘密泄露風(fēng)險，企業(yè)可借鑒日本終身雇傭模式，通過(guò)構建長(cháng)期信任關(guān)系降低泄密隱患。在技術(shù)獲取路徑上，齊寶鑫主張多元化策略，將自主研發(fā)的深度與校企合作、成果轉化的廣度相結合，以提升效率并分散風(fēng)險。針對AI技術(shù)對知識產(chǎn)權法的沖擊，他提出應以“創(chuàng )意參與度”作為獨創(chuàng )性判定標準，同時(shí)呼吁法律框架保持開(kāi)放性與適應性。此外，齊寶鑫以國產(chǎn)化C919客機為例，指出自主創(chuàng )新是突破國際技術(shù)壁壘的核心路徑；而在高風(fēng)險研發(fā)管理中，他建議依托商業(yè)秘密確權機制與哈希值技術(shù)，強化核心數據保護，尤其需警惕跨國合作中的技術(shù)外溢風(fēng)險?！?/b>

來(lái)源：IPRdaily中文網(wǎng)（iprdaily.cn）

2025年5月18日，由復旦大學(xué)管理學(xué)院、華盛頓大學(xué)麥克道爾國際學(xué)者研究院、北京大學(xué)光華管理學(xué)院、華盛頓大學(xué)奧林商學(xué)院聯(lián)合主辦的第三屆全球創(chuàng )新與知識產(chǎn)權論壇在中國上海復旦大學(xué)管理學(xué)院圓滿(mǎn)落幕。齊寶鑫博士代表IPR Daily作為嘉賓受邀參會(huì )，并在“商界領(lǐng)袖的聲音”環(huán)節分享前沿見(jiàn)解。

附：發(fā)言全文（雙語(yǔ)）

Q：您認為當今知識產(chǎn)權戰略面臨的主要挑戰和趨勢有哪些？

What are the key challenges and trends you see today in IP strategy？

A：基于我的工作經(jīng)驗，當下知識產(chǎn)權戰略面臨的挑戰之一就是較高的人員流動(dòng)性帶來(lái)的知識產(chǎn)權保護問(wèn)題。這一挑戰也反映出當下知識產(chǎn)權戰略調整的一個(gè)趨勢就是商業(yè)秘密的重要性愈發(fā)凸顯。在一個(gè)企業(yè)中研發(fā)總監和銷(xiāo)售總監總是能有一份不錯的薪資待遇，因為他們通常掌握著(zhù)一個(gè)企業(yè)的技術(shù)秘密和經(jīng)營(yíng)秘密。如何保護這些商業(yè)秘密是隨之而來(lái)的一個(gè)重要問(wèn)題。日本傳統的終身雇傭制度，為商業(yè)秘密的保護提供了極具參考價(jià)值的實(shí)踐樣本。這一制度基于長(cháng)期勞動(dòng)合同關(guān)系，構建起企業(yè)與員工之間穩定的法律契約，員工自入職起便與企業(yè)形成貫穿職業(yè)生命周期的權利義務(wù)關(guān)系。再沒(méi)有比一個(gè)工程師東京大學(xué)畢業(yè)就進(jìn)入豐田公司從事研發(fā)，一直到退休，從未進(jìn)入過(guò)豐田競爭對手的公司，更能保護豐田的商業(yè)秘密了。從知識產(chǎn)權保護視角審視，該制度呈現出三重法律優(yōu)勢：首先，員工忠誠度高，泄密風(fēng)險低。長(cháng)期服務(wù)于同一企業(yè)的員工更容易形成歸屬感和責任感，他們對公司機密信息的保護意識也更強，從而減少了故意泄密或“跳槽帶走技術(shù)”的風(fēng)險。其次，知識流動(dòng)受控，保密鏈條更穩定。相比頻繁更換工作的環(huán)境，終身雇傭減少了核心技術(shù)和商業(yè)信息在行業(yè)內的頻繁流動(dòng)，保護了企業(yè)的研發(fā)成果和戰略布局。最后，企業(yè)更愿意投入培訓和信任。由于員工不會(huì )輕易離職，企業(yè)可以放心將關(guān)鍵技術(shù)、管理經(jīng)驗傳授給員工，這在加強團隊能力的同時(shí)，也建立了更緊密的信任關(guān)系，從制度上降低了因信息外泄而造成的損失。這種終身雇傭制度能夠有效降低人員流動(dòng)這一商業(yè)秘密泄露主要原因帶來(lái)的風(fēng)險。當然，終身雇傭制并不一定完全適合國內環(huán)境，但是它提供了一個(gè)降低商業(yè)秘密泄露風(fēng)險的思路，可以作為我們知識產(chǎn)權保護的一個(gè)重要參考。

From my work experience, one of the key challenges that IP strategies are facing is the issue of IP protection arising from high workforce mobility. This challenge also reflects an emerging trend in contemporary IP strategy adjustments - the increasing importance of trade secrets. Within enterprises, R&D Directors and Sales Directors usually can get high salaries, mainly because they have a knowledge of a company’s technical secrets and business secrets. How to protect these trade secrets has consequently become a critical issue.Japan's traditional lifetime employment system offers an instructive model for trade secret protection. This system establishes stable legal contractual relationships through long-term employment, creating career-long rights and obligations between enterprises and employees from recruitment to retirement. There is nothing more effective in protecting Toyota's trade secrets than having an engineer who, upon graduating from the University of Tokyo, joins Toyota's R&D department and stays there till retirement, without ever transitioning to any Toyota's competitors.From an IP protection perspective, this system demonstrates three legal advantages: Firstly, employee loyalty will be high, and the risk of information leakage will be low. Employees who have served the same enterprise for a long time are more likely to develop a sense of belonging and responsibility. They also have a stronger awareness of protecting the company's confidential information, thus reducing the risk of deliberate leakage of secrets or taking technologies when changing jobs. Secondly, the spread of knowledge will be controlled, and the confidentiality chain will be more stable. Compared with an environment where employees change jobs frequently, lifetime employment reduces the frequent dissemination of core technology and business information within industries, preserving R&D achievements and strategic layouts of the company. Thirdly, the company will be more willing to invest in training and place trust. Since employees are not likely to leave the company, the company can confidently impart core technologies and management experience to them. This not only strengthens the team's capabilities but also establishes a closer relationship of trust, reducing the losses caused by information leakage from an institutional perspective.This lifetime employment system can effectively reduce the risk of trade secret leakage caused by high employee turnover, which is one of the primary contributing factors. While such lifetime employment system may not fully align with domestic contexts, it provides a conceptual framework for reducing trade secret leakage risks and serves as a valuable reference for enhancing intellectual property protection.

Q：企業(yè)未來(lái)幾年成功應對挑戰的關(guān)鍵要素是什么？

What do you see as the key factors for enterprises to successfully meet the challenges in the coming years?

A：掌握核心技術(shù)是企業(yè)保持生命力的關(guān)鍵。當想到獲取技術(shù)時(shí)人們通常第一反應是通過(guò)自主研發(fā)獲取技術(shù)，我將它比喻成“自建房”。但是在“自建房”以外還有買(mǎi)房、租房甚至合建、合買(mǎi)、合租的方式。這些方式各有利弊，對于自建房來(lái)說(shuō)能夠做到技術(shù)與需求達到最大匹配度，具有良好的效益；但是自建房要從地基開(kāi)始、從一磚一瓦開(kāi)始，周期長(cháng)效率低。相反，租房模式效率高，能夠很快就獲取需要的技術(shù)，但不能保證“租期”有多久，說(shuō)不定明天房東就不愿意出租房屋了。而買(mǎi)房模式能夠很快獲取技術(shù)而不用擔心房子會(huì )被房東收走，問(wèn)題在于可能無(wú)法保證買(mǎi)來(lái)的房子將來(lái)不會(huì )貶值。而與他人合作，合建、合買(mǎi)、合租更能提高效率、降低成本，但合作方可能會(huì )帶來(lái)風(fēng)險。同時(shí)不要忘記大學(xué)和科學(xué)研究所通常擁有比較多的技術(shù)，與他們合作或者向他們“租房”“買(mǎi)房”是個(gè)不錯的選擇。2015年修訂的促進(jìn)科技成果轉化法也有效促進(jìn)了與大學(xué)或者科研院所之間的合作。有人可能擔心大學(xué)或者科研院所的技術(shù)會(huì )在最后一公里上出問(wèn)題，這種情況下企業(yè)在選擇合作院所時(shí)則需要提高判斷能力，必要時(shí)候委托專(zhuān)業(yè)人員給出專(zhuān)業(yè)的判斷意見(jiàn)。

Mastering core technologies is crucial for enterprises to maintain vitality. When it comes to acquiring technologies, people usually first think of independent research and development, which I metaphorically call "self-built house". However, besides "self-built house", there are alternative approaches like "purchasing", "renting", or even "co-building", "co-purchasing", and "co-renting". Each approach has its pros and cons.Self-built enables maximum alignment between technology and needs with good returns, but requires starting from laying foundations brick by brick, resulting in long cycles and low efficiency. In contrast, the "rental model" offers high efficiency for quick technology access, but cannot guarantee lease duration - the "landlord" might terminate the agreement at any time. The "purchase model" allows rapid technology acquisition without repossession concerns, but carries risks of technological depreciation. Collaborative approaches like co-development, joint procurement, and shared leasing can improve efficiency and reduce costs, though partners may introduce risks.Notably, universities and research institutions often possess substantial technological resources. Collaborating with them or "renting/buying" technologies from them will also be a good choice. The Law on Promoting the Transformation of Scientific and Technological Achievements amended in 2015 has effectively facilitated such collaborations. While some worry about potential "last mile" issues in academic technological achievements, enterprises should enhance their discernment capabilities when selecting partners, and seek professional evaluations when necessary.

Q：在此新背景下，我們是否需要重新定義對知識產(chǎn)權及其保護的理解？還是既有原則依然適用？

Do we need to re-think our understanding of IP and IP protection in this new context or do the same principles apply?

A：新背景下面對AI等新技術(shù)帶來(lái)的知識產(chǎn)權新問(wèn)題，我們應當持開(kāi)放的態(tài)度。我們都知道，法律具有一定的滯后性，面對新出現的AI等技術(shù)帶來(lái)的問(wèn)題時(shí)，現有的知識產(chǎn)權法難以完全應對，我們更多的是在已有的法律上通過(guò)各種解釋方法來(lái)適應這些問(wèn)題。但是我們也看到了，現有法律無(wú)法有效解決AI等技術(shù)帶來(lái)的每一個(gè)問(wèn)題，我們暫時(shí)也無(wú)法完全預知AI這些技術(shù)將來(lái)的走向和將來(lái)可能帶來(lái)的問(wèn)題。2023年北京互聯(lián)網(wǎng)法院判決的第一例AI文生圖引發(fā)了很大爭議。該案中，法院認為AI文生圖可以獲得版權法保護，圖片的創(chuàng )作過(guò)程體現了作者的獨創(chuàng )性。然而問(wèn)題在于，通過(guò)向AI輸入關(guān)鍵詞進(jìn)行創(chuàng )作，如何確定獨創(chuàng )性出現的界限呢？輸入十個(gè)關(guān)鍵詞、二十個(gè)關(guān)鍵詞、還是更多關(guān)鍵詞，進(jìn)行一次調整、兩次調整還是更多次調整能體現出作者的獨創(chuàng )性呢？這些都難以得出一個(gè)明確的結論。關(guān)于這一點(diǎn)，在日本知識產(chǎn)權戰略本部的報告書(shū)中提出，當AI使用者向已訓練完成的AI模型輸入所選擇的圖像，或從大量生成的AI產(chǎn)出物中挑選出多個(gè)進(jìn)行公開(kāi)發(fā)布的情況下，如果使用者在選擇或構成等方面體現出一定的創(chuàng )意與工夫，其參與行為本身有可能被認定為具有創(chuàng )作性。但僅僅設置參數，則不能能說(shuō)是創(chuàng )作。因此，對于A(yíng)I這些新技術(shù)或許我們不需要急著(zhù)訂立新的法律去給他們框出一個(gè)發(fā)展的范圍，先讓子彈飛一會(huì )，看看他們會(huì )去向何方。就像當初互聯(lián)網(wǎng)的誕生帶來(lái)的版權問(wèn)題一樣，互聯(lián)網(wǎng)的出現在傳統的紙媒發(fā)表以外提供了信息網(wǎng)絡(luò )發(fā)表的途徑，從而帶來(lái)了信息網(wǎng)絡(luò )傳播權。也許在我們對AI這些新技術(shù)更熟悉一些以后能夠發(fā)現更合適他們的法律保護方式。

Under the new landscape where emerging technologies like AI pose novel intellectual property challenges, we should maintain an open mindset. As we recognize, legal frameworks inherently lag behind technological advancements. When confronting AI-related issues, existing IP laws struggle to provide comprehensive solutions, we mainly use various interpretive methods to adapt to these issues based on the existing laws. However, it's evident that current legislation cannot fully address every challenge posed by AI, nor can we entirely predict its future trajectory and associated complications.The landmark 2023 Beijing Internet Court ruling on AI-generated text-to-image works sparked significant controversy. In this case, the court recognized AI-generated images as copyrightable works, acknowledging the human creator's originality in the generation process. Nevertheless, critical questions persist: Where should we draw the line for originality in works created through keyword inputs to AI systems? Does originality emerge from inputting ten keywords versus twenty? How many iterative adjustments constitute sufficient creative input? These thresholds remain ambiguous. Regarding this matter, the Intellectual Property Strategy Headquarters of Japan stipulated in its official report: When AI users either input selected images into pre-trained AI models or curate multiple AI-generated outputs from extensive results for public release, their participatory actions may be recognized as constituting creative contributions if their selection or compositional arrangements demonstrate intentional creative input. However, mere parameter configuration would not qualify as creative authorship.Therefore, perhaps we need not rush to establish restrictive new legislation for AI technologies. Instead, we might "let the bullet fly for a while," allowing space to observe their natural evolution. Historically, the internet's emergence created new copyright dimensions beyond traditional print media, ultimately leading to the recognition of information network dissemination rights. Similarly, as we deepen our understanding of AI's capabilities and implications, we may develop more nuanced legal protections tailored to these technologies' unique characteristics. This adaptive approach enables legal frameworks to mature alongside technological progress rather than prematurely constraining innovation.

Q：面對全球創(chuàng )新活動(dòng)布局中的相互依賴(lài)性與潛在風(fēng)險，本土創(chuàng )新的合理邊界應如何界定？

In light of the global interdependencies and associated vulnerabilities in considering where to locate innovation activities, what is the right level of indigenous innovation?

A當下正處于中美脫鉤的國際環(huán)境，在全球技術(shù)創(chuàng )新活動(dòng)中我們面臨著(zhù)美國技術(shù)限制的問(wèn)題，對于本土創(chuàng )新最好的做法就是加大自主研發(fā)、鼓勵自研，盡一切努力做到我們能做的。實(shí)現品牌和技術(shù)上的自主。最好的例子就是中國民航發(fā)展的ABC，從購買(mǎi)空客、波音的飛機到我們C919自主研發(fā)率達60%，主要核心部件已經(jīng)實(shí)現自主研發(fā)，最有力地證明了自主創(chuàng )新是當下國際形式下的最優(yōu)選擇。

Amid the current international landscape of Sino-US decoupling, where global technological innovation faces increasing American technical restrictions, the optimal approach for indigenous innovation lies in intensifying independent R&D efforts and encouraging self-developed technologies. We must exhaust all efforts to achieve brand autonomy and technological self-sufficiency. A compelling example is the evolution of China's civil aviation sector – transitioning from purchasing Airbus and Boeing aircraft to developing the C919 with a 60% localization rate, where core components now demonstrate independent R&D capabilities. This progression powerfully validates that autonomous innovation constitutes the optimal choice under current geopolitical dynamics.

Q：除專(zhuān)利申請外，企業(yè)在跨國/國內高風(fēng)險研發(fā)過(guò)程中可采取哪些策略工具？

Besides filing for patents, what are the tools businesses can use to navigate the risky R&D processes, both within and across countries?

A：我們還可以選擇以商業(yè)秘密方式保護知識產(chǎn)權。在某些程度上，商業(yè)秘密在知識產(chǎn)權保護中比專(zhuān)利發(fā)揮更重要的作用，技術(shù)空殼公司所持有的發(fā)明專(zhuān)利并不重要，技術(shù)人員掌握的商業(yè)秘密才重要。一些公司會(huì )假借并購的名義，挖走研發(fā)負責人及其帶領(lǐng)的團隊，并進(jìn)一步獲得了目標公司的全部的技術(shù)秘密和經(jīng)營(yíng)信息。通過(guò)公證購買(mǎi)產(chǎn)品，繼而進(jìn)行反向工程研究，可以判斷是否侵犯技術(shù)秘密?，F在，上海技術(shù)交易所已經(jīng)具備了提供商業(yè)秘密確權的服務(wù)能力。簡(jiǎn)單來(lái)說(shuō)，這項服務(wù)讓企業(yè)或者科研機構可以把一些不方便公開(kāi)、也不適合申請專(zhuān)利的技術(shù)成果，通過(guò)一種“非公開(kāi)”的方式登記下來(lái)，在技術(shù)交易、融資，或者發(fā)生法律糾紛的時(shí)候，都可以用來(lái)證明權屬。尤其對于A(yíng)I這些新技術(shù)，在利用專(zhuān)利進(jìn)行保護仍存在較大困難時(shí)可以考慮采用商業(yè)秘密的方式進(jìn)行保護。比如AI模型的參數、訓練數據、核心算法這些東西，只要企業(yè)有實(shí)用價(jià)值、沒(méi)公開(kāi)過(guò)，并且采取了保密措施，就可以通過(guò)確權，把它們作為商業(yè)秘密登記保護起來(lái)。在此過(guò)程中，有一個(gè)很關(guān)鍵的技術(shù)叫做哈希值。它像是一種“數字指紋”——可以幫我們給某個(gè)AI模型或生成內容生成一個(gè)獨一無(wú)二的代碼。哈希值的利用不僅提高了確權的效率和可信度，還能幫助我們在后續交易、安全共享、甚至維權時(shí)，提供強有力的技術(shù)支撐。

We may also opt for trade secret protection in intellectual property management. To certain extent, trade secrets play a more critical role than patents in technological competition - the invention patents held by patent shell companies hold little value, whereas the trade secrets mastered by technical personnel constitute the real competitive advantage. Some corporations strategically acquire R&D leadership and their teams under the guise of mergers and acquisitions, thereby obtaining the target company's complete technical know-how and operational intelligence. Through notarized product acquisition followed by reverse engineering analysis, one can effectively determine whether technical secret infringement has occurred.The Shanghai Technology Exchange has now operationalized trade secret verification services. Essentially, this initiative enables enterprises and research institutions to register non-disclosable technological achievements—those unsuitable for patent applications—through confidential filing mechanisms. Such registration serves as legally-recognized ownership evidence during technology transfers, financing activities, or legal disputes. Particularly for emerging technologies like AI where patent protection remains challenging, trade secret mechanisms present viable alternatives. Critical AI components including model parameters, training datasets, and core algorithms can be registered as protected trade secrets, provided they maintain commercial value, remain undisclosed, and implement proper confidentiality protocols.A pivotal technological enabler in this process is hash value application. Functioning as a "digital fingerprint", hash algorithms generate unique cryptographic identifiers for AI models or generated content. This technological implementation not only enhances verification efficiency and credibility but also provides robust technical support throughout subsequent transactions, secure data sharing, and rights enforcement processes.

（原標題：齊寶鑫：淺談當前形勢下的知識產(chǎn)權戰略面臨的主要挑戰和趨勢）

來(lái)源：IPRdaily中文網(wǎng)（iprdaily.cn）

編輯：IPRdaily辛夷          校對：IPRdaily縱橫君

注：原文鏈接：齊寶鑫：淺談當前形勢下的知識產(chǎn)權戰略面臨的主要挑戰和趨勢（點(diǎn)擊標題查看原文）

「關(guān)于IPRdaily」

IPRdaily是全球領(lǐng)先的知識產(chǎn)權綜合信息服務(wù)提供商，致力于連接全球知識產(chǎn)權與科技創(chuàng )新人才。匯聚了來(lái)自于中國、美國、歐洲、俄羅斯、以色列、澳大利亞、新加坡、日本、韓國等15個(gè)國家和地區的高科技公司及成長(cháng)型科技企業(yè)的管理者及科技研發(fā)或知識產(chǎn)權負責人，還有來(lái)自政府、律師及代理事務(wù)所、研發(fā)或服務(wù)機構的全球近100萬(wàn)用戶(hù)（國內70余萬(wàn)+海外近30萬(wàn)），2019年全年全網(wǎng)頁(yè)面瀏覽量已經(jīng)突破過(guò)億次傳播。

（英文官網(wǎng)：iprdaily.com  中文官網(wǎng)：iprdaily.cn） 

本文來(lái)自IPRdaily中文網(wǎng)（iprdaily.cn）并經(jīng)IPRdaily.cn中文網(wǎng)編輯。轉載此文章須經(jīng)權利人同意，并附上出處與作者信息。文章不代表IPRdaily.cn立場(chǎng)，如若轉載，請注明出處：“http://qaqi.cn”