Artificial Intelligence And Patent Law - An Analysis Of Inventorship And Ownership Through The DABUS Litigation Framework
Author: LawSphereX
Category: Technology, Cyber & Privacy
Published:
Updated:
ARTIFICIAL INTELLIGENCE AND PATENT LAW: AN
ANALYSIS
OF INVENTORSHIP AND OWNERSHIP THROUGH THE DABUS LITIGATION FRAMEWORK Riya Jaiswal, Student, Department of Industrial Engineering and Management (IEM), R. V College of Engineering (RVCE), Bengaluru Anchal Karanani, Student, Department of Industrial Engineering and Management (IEM), R. V College of Engineering (RVCE), Bengaluru Shree Maheshwari, Student, Department of Industrial Engineering and Management (IEM), R. V College of Engineering (RVCE), Bengaluru Chitra B. T. , Assistant Professor, Department of Industrial Engineering and Management (IEM), R. V College of Engineering (RVCE), Bengaluru inventor. Patent authorities in United Kingdom, European Patent Office, United States, and additional jurisdictions rejected these applications, confirming human-only inventorship requirements.
Judicial decisions, notably the 2023 UK Supreme Court judgment and 2022 US Court of Appeals for Federal Circuit opinion, established doctrinal positions while leaving unresolved questions regarding ownership structures, incentive mechanisms, and patent system adaptation for autonomous artificial intelligence. Abstract—Contemporary artificial intelligence systems demonstrate capabilities for generating technical solutions with reduced human intervention, thereby challenging established patent law principles requiring natural persons as inventors. This research examines inventorship and ownership challenges when artificial intelligence contributes substantially to patentable innovations.
A comparative examination of DABUS patent applications across United Kingdom, European Patent Office, and United States jurisdictions, combined with evaluation of United States Patent and Trademark Office guidance documents from 2024 and 2025, forms the methodological foundation. The investigation addresses why patent frameworks mandate human inventors, ownership allocation mechanisms involving artificial intelligence, and policy implications of alternative regulatory approaches. Through doctrinal legal methodology and comparative jurisprudence analysis, four regulatory frameworks are evaluated: preserving current human-centric models, implementing statutory deeming mechanisms, establishing limited legal personhood, and creating specialized intellectual property regimes.
Assessment criteria include innovation promotion, accountability frameworks, administrative feasibility, and international harmonization prospects. Application of this analytical framework to India’s Patents Act, 1970 identifies regulatory gaps and recommends legislative modifications. Findings reveal fundamental tensions between doctrinal consistency and technological accommodation, with implications for global patent systems. A. Research Motivation Patent law fulfills dual objectives: compensating inventors for knowledge disclosure and incentivizing continued innovation. Both objectives presume human agency. When artificial intelligence systems generate inventions, traditional doctrines encounter four interconnected challenges:
1 Inventorship Attribution
When conception occurs within artificial intelligence processing, can humans who trained, prompted, or deployed the system claim inventorship status?
2 Ownership Allocation
If artificial intelligence cannot qualify as inventor, which parties own patent rights in artificial intelligence-generated inventions? 3) Disclosure Requirements: How should applicants accurately represent artificial intelligence contributions without invalidating applications?
4 Policy Coherence
Which incentive structures optimally promote artificial intelligence innovation while maintaining accountability? These challenges possess practical significance as pharmaceutical and materials research increasingly employs generative computational models, requiring inventors and patent practitioners to determine appropriate artificial intelligence contribution characterization. Index Terms—Artificial intelligence, patent law, inventorship, ownership, DABUS litigation, comparative legal analysis, USPTO guidance, Indian Patents Act I.
INTRODUCTION
A RTIFICIAL intelligence evolution has transitioned from computational assistance tools to systems capable of autonomous technical problem-solving. Contemporary machine learning algorithms design pharmaceutical compounds, optimize structural engineering solutions, and develop novel materials with minimal direct human cognitive involvement in creative processes. This technological progression challenges patent law foundations premised on natural persons as invention creators. B. Research Questions This investigation addresses four primary questions: RQ1: Under existing patent frameworks, what doctrinal and policy rationales support limiting inventorship to natural persons? The intersection of artificial intelligence capabilities and patent doctrine became prominent through patent applications submitted by Dr.
Stephen Thaler designating DABUS (Device for Autonomous Bootstrapping of Unified Sentience) as sole Intermediate scenarios include AI-assisted invention where humans use artificial intelligence to explore design spaces while retaining control over problem definition and solution selection. AI-augmented invention involves artificial intelli- AI- Augmented AI as AI- Assisted AI- Generated Tool AI Autonomy gence proposing solutions that humans evaluate and refine with balanced contributions. AI-generated invention occurs when artificial intelligence autonomously produces inventive concepts with human roles limited to recognizing utility and filing applications. Clear human conception Inventorship uncertainty Legal Implication Threshold Fig. 1.
AI Contribution Spectrum: From tool to autonomous inventor, with legal implications demarcated by the threshold where human conception becomes uncertain. This spectrum matters legally because patent systems require inventor identification based on conception. If artificial intelligence contributions cross thresholds where artificial intelligence, rather than humans, performs conception mental acts, traditional inventorship doctrine experiences strain. RQ2: When artificial intelligence substantially contributes to or autonomously generates inventions, who should hold patent rights and through which legal mechanisms? RQ3: What policy trade-offs exist among different legal approaches regarding innovation incentives, accountability, disclosure integrity, and litigation risk? B. Patent Law Fundamentals Patent law distinguishes inventorship from ownership.
Inventorship constitutes factual determination of who conceived the invention. Ownership represents legal determination of who holds exclusionary rights. RQ4: Which lessons from DABUS litigation and USPTO guidance apply to India under Patents Act, 1970, particularly considering requirements for true and first inventors? 1) Inventorship: An inventor contributes to claimed invention conception. Conception represents definite and permanent idea formation in inventor’s mind of complete and operative invention. Conception constitutes mental act distinct from reduction to practice or problem recognition. In collaborative inventions, co-inventors must contribute significantly to conception without requiring equal contributions. C.
Research Contributions This work provides: • Comprehensive comparative analysis of DABUS litigation outcomes across UK, EPO, and US jurisdictions • Original taxonomy of policy alternatives for artificial intelligence inventorship and ownership with criteriabased evaluation • Integration of USPTO 2024 and 2025 guidance into comparative framework • India-specific analysis applying comparative framework to Patents Act, 1970 • Visual analytical tools for practitioners and policymakers Patent statutes historically assume natural person inventors. US Patent Act references inventor or inventors and whoever invents. European Patent Convention Article 60(1) states rights belong to inventor or successor in title. UK Patents Act 1977 section 7 provides patent grants to inventor or title-deriving persons. India’s Patents Act, 1970 section 6 allows applications by true and first inventor.
2) Ownership: Patent ownership may differ from inventorship through assignment, employment agreements, or statutory provisions. In United States, inventors initially own inventions but may assign rights to employers or other parties. Workfor-hire doctrines can vest ownership in employers when employees invent within employment scope. In India, section 6 permits applications by true and first inventor or title-deriving persons, with employer ownership recognized under section 6(b) for employment-course inventions. Assignment requires transfer from inventor to another person. If artificial intelligence systems constitute actual inventors, no human inventors exist from whom to assign rights, creating legal gaps. D. Paper Structure Section II establishes background on artificial intelligence invention types and patent fundamentals. Section III reviews related scholarship.
Section IV describes legal frameworks in key jurisdictions. Section V presents DABUS case chronology. Section VI conducts comparative analysis. Section VII evaluates policy options. Section VIII applies framework to India. Section IX concludes with future research directions. II.
BACKGROUND
A. Artificial Intelligence Contribution Spectrum Artificial intelligence participation in invention generation exists along a continuum, illustrated in Fig. 1. At one extreme, artificial intelligence serves as passive computational tool where human inventors formulate hypotheses, design experiments, and interpret results while artificial intelligence performs calculations. At the opposite extreme, artificial intelligence systems autonomously identify problems, generate candidate solutions, evaluate alternatives, and output novel inventions with minimal human direction beyond initial training. C. Tool Theory Versus Agency A central conceptual question addresses whether artificial intelligence should be characterized as tool wielded by human inventors or agent capable of inventive activity.
Under tool theory, artificial intelligence analogizes to microscopes, computers, or laboratory instruments aiding human inventors without conceiving. Under agency theory, sufficiently advanced artificial intelligence could perform conception mental acts, inventions distinct from patents [17]. However, sui generis systems raise coordination problems for international treaties. TABLE I TAXONOMY OF ARTIFICIAL INTELLIGENCE INVENTORSHIP SCHOLARSHIP Type Focus and Scope C. Empirical and Interdisciplinary Work Doctrinal Analysis Interpretation of existing statutes and case law; DABUS decision analysis; inventorship requirement comparison across jurisdictions Empirical research on artificial intelligence impact on patenting remains nascent. Studies survey patent filings mentioning artificial intelligence or machine learning, finding rapid growth [18].
Others model economic effects of different ownership rules [19]. Interdisciplinary work brings computer Policy & Normative Proposals for legal reform including sui generis rights, deeming provisions, personhood; normative arguments on incentive theory and accountability science perspectives on autonomy and creativity [20]. Empirical & Interdisciplinary Surveys of artificial intelligence role in research and development; economic modeling of incentive effects; computer science perspectives on autonomy IV. LEGAL FRAMEWORKS: COMPARATIVE OVERVIEW A. United States
1 Inventorship
US patent law derives from Constitutional Patent Clause and is codified in Title 35 US Code. Section
101 provides whoever invents or discovers may obtain patents.
Section 115 requires applications include inventor oaths, identifying inventor and affirming original inventor status. Case law defines conception as definite idea formation of complete and operative invention, a mental act performed by natural persons [12]. potentially qualifying as inventor if legal personhood were recognized. Most jurisdictions currently adhere to tool theory, treating artificial intelligence as sophisticated instrument. This approach maintains doctrinal consistency but may experience strain as artificial intelligence capabilities advance toward autonomous problem formulation and solution. In Thaler v. Vidal, US Court of Appeals for Federal Circuit affirmed that inventor under Patent Act requires natural person [7]. The court reasoned statute’s use of whoever and individual, along with oath requirements signed by inventor, presuppose human inventors. III.
RELATED WORK Scholarship on artificial intelligence and patent law has expanded significantly since 2019 when DABUS applications attracted attention. Literature classifies into three categories as shown in Table I.
2 Ownership
Inventors initially own inventions but may assign rights to others. Employment agreements often require employees to assign inventions made within employment scope. When artificial intelligence generates inventions, if humans are deemed inventors, standard assignment mechanisms apply. If no human qualifies as inventor, ownership gaps emerge. A. Doctrinal Analysis World Intellectual Property Organization produced extensive analysis identifying inventorship and ownership as central challenges, noting patent systems were designed without contemplating non-human inventors [1]. Subsequent publications surveyed national approaches and synthesized policy considerations [2], [3].
3 USPTO Guidance
In February 2024, USPTO issued guidance clarifying inventorship for artificial intelligenceassisted inventions [9]. Guidance reaffirmed only natural persons may be inventors but acknowledged artificial intelligence use does not automatically disqualify persons from inventorship. Key inquiry addresses whether natural persons made significant contributions to claimed invention conception. EPO decision in J 8/20 analyzed in guidance and press releases concluded European Patent Convention designation requirement presumes natural person inventors [4], [5]. Academic commentators examined DABUS decisions focusing on statutory construction, policy rationales, and procedural issues [6].
November 2025 revised guidance provided additional examples clarifying that persons designing artificial intelligence systems with capability to generate inventive concepts may be inventors if persons made significant contributions to ultimately claimed inventions [10]. Merely presenting problems to artificial intelligence and recognizing useful output does not constitute inventorship. B. Policy and Normative Proposals Normative scholarship proposes various reforms. Some argue for maintaining human-only inventorship, emphasizing accountability and doctrinal stability [13], [14]. Others propose deeming provisions where humans controlling or prompting artificial intelligence could be deemed inventors under defined conditions [15]. B. United Kingdom A minority advocates limited legal personhood for artificial intelligence systems, analogizing to corporate personhood [16].
Critics argue this would undermine incentive theory and create enforcement difficulties. 1) Inventorship and Ownership: UK Patents Act 1977 implements European Patent Convention in domestic law. Section
7 specifies who may be granted patents: inventor or successor
in title, employer in certain cases, or person entitled under foreign law. Section 13 requires applications state inventor. Sui generis rights are proposed as alternatives, creating new protection categories for artificial intelligence-generated
3 Ownership and Entitlement
Section 6 provides routes for non-inventors to obtain patents: assignees and legal representatives. If artificial intelligence generates inventions, human applicants might argue entitlement as assignee of artificial intelligence owner. However, this requires artificial intelligence could hold rights initially, which India’s legal system does not recognize. UK Intellectual Property Office rejected DABUS application, holding Patents Act reference to inventor and requirements for providing inventor name presuppose human inventors. On appeal, UK Supreme Court in Thaler v. Comptroller- General of Patents held inventor must be person, meaning natural person, and artificial intelligence cannot be named as inventor [8]. The Court reasoned Patents Act provisions, including requirements to state inventor name and address and provisions for inventor compensation, assume human inventors.
The Court noted recognizing artificial intelligence as inventor would require legislative amendment. Employer ownership under section 6(b) might apply if employees use artificial intelligence in employment course. Employees could be deemed inventors if they made significant contributions, and employers would hold rights. V. DABUS LITIGATION TIMELINE C. European Patent Office Fig. 2 presents chronological development of DABUSrelated events across jurisdictions.
1) Legal Framework
European Patent Convention governs European patent applications. Article 60(1) provides rights to European patents belong to inventor or successor in title. Rule
19 requires inventor designation in applications.
EPO Guidelines for Examination state inventors must be natural persons [6]. Inventor designation serves to establish who conceived invention and supports entitlement determination. A. Initial Filings and Rejections Dr. Stephen Thaler filed patent applications in multiple jurisdictions designating DABUS, described as creativity machine, as sole inventor. Thaler stated he owned DABUS and derived entitlement to invention from ownership and creation of DABUS.
2 Decision in J 8/20
In DABUS application EP 18 275 163, applicant designated DABUS as inventor stating entitlement derived from being owner and creator of DABUS. EPO refused application for failure to designate inventor in accordance with Rule 19 EPC. UKIPO refused application in December 2019, holding Patents Act requires inventor be natural person. EPO issued similar refusal in December 2019, finding inventor designation did not comply with EPC requirements. USPTO rejected applications in April 2020, determining Patent Act requirements presuppose human inventors. Applicant appealed to Legal Board of Appeal, which issued decision J 8/20 [4].
Board held EPC designation requirement presupposes inventor is natural person, reasoning EPC language implies human characteristics, designation serves to identify inventive thought source which only humans provide, and allowing artificial intelligence as inventor would undermine designation purpose and create inconsistencies with national laws. B. Appeals and Judicial Decisions
1 EPO Legal Board
Thaler appealed EPO refusal to Legal Board of Appeal. In decision J 8/20 issued December 2021, Board dismissed appeal [4]. Board held EPC requires designation of inventor who is natural person, purpose of inventor designation is identifying person whose mental activity led to invention, machines lack legal personhood and cannot perform legal acts associated with inventorship, and argument that denying artificial intelligence inventorship leaves inventions unprotected is policy concern for legislators. D. India
1 Patents Act, 1970
India’s patent system is governed by Patents Act, 1970 [11]. Section 6 specifies who may apply for patents: true and first inventor, person being assignee of true and first inventor, or legal representative of deceased inventor or assignee.
2 US Court of Appeals
Thaler appealed USPTO rejection to US District Court for Eastern District of Virginia, which granted summary judgment to USPTO. Thaler then appealed to Court of Appeals for Federal Circuit. Section 6(b) clarifies that for inventions made by employees in employment course, employers may apply if invention relates to employer business and was made using employer resources. In August 2022, CAFC issued opinion affirming district court [7]. Court applied statutory construction canons and held Patent Act use of whoever in section 101 and individual in section 115 refers to natural persons, not machines. Act requires inventor oath which presupposes human agency. Legislative history and predecessor statutes consistently assumed human inventors. Expanding inventorship to artificial intelligence would require clear congressional authorization.
Act does not explicitly define inventor, but section 2(1)(y) defines true and first inventor excluding persons who import inventions or derive inventions from others without substantial contribution. This negative definition suggests inventor must be person capable of legal relationships and moral rights. 2) Current Position: India has not faced DABUSequivalent cases, and neither Patent Office nor courts have issued formal guidance on artificial intelligence inventorship. However, doctrinal analysis suggests India likely requires inventors be natural persons. Requirement to state true and first inventor and provisions for inventor compensation presuppose human inventors.
3 UK Supreme Court
Thaler appealed UKIPO refusal through UK courts, reaching Supreme Court. In December 2023, Supreme Court dismissed appeal [8]. Court held Patents Act 1977 requires inventor be natural person as evidenced by references to inventor name and address, inventor compensation, and derivation of title from inventor. Artificial intelligence 2018-2019: Thaler files applications in EPO, UKIPO, USPTO naming DABUS Dec 2019: EPO refuses - improper inventor designation Dec 2019: UKIPO refuses - inventor must be natural person Apr 2020: USPTO rejects - DABUS cannot be inventor Jul 2021: South Africa grants patent Dec 2021: EPO Board issues J 8/20 upholding refusal Aug 2022: US Federal Circuit affirms in Thaler v. Vidal Dec 2023: UK Supreme Court dismisses appeal Feb 2024: USPTO issues AI-Assisted Inventions Guidance Nov 2025: USPTO issues Revised Guidance Fig. 2.
Timeline of DABUS litigation and related guidance developments (2018-2025). VI.
COMPARATIVE
ANALYSIS
Table II summarizes jurisdictional approaches across key dimensions. cannot be inventor because it lacks legal personhood and cannot hold rights. Ownership of artificial intelligence system does not automatically confer entitlement to its inventions because entitlement requires derivation from inventor who is person. A. Inventorship Tests All examined jurisdictions require inventors be natural persons. This consensus rests on multiple rationales:
1 Statutory Language
Patent statutes use terms courts interpret as referring to humans. US Patent Act requirement for inventor oath which must be signed and sworn presupposes human agency. UK Patents Act provisions for inventor compensation and moral rights assume human inventors. EPC designation requirement serves to identify human source of inventive thought. C. Guidance Developments In response to growing artificial intelligence use in invention, USPTO issued formal guidance in February 2024 [9]. Guidance clarified only natural persons can be inventors but use of artificial intelligence tools does not preclude person from being inventor if person made significant contribution to conception.
2 Conception as Mental Act
Inventorship doctrine centers on conception—definite and permanent idea of complete invention formed in inventor mind. This mental act is inherently human. While artificial intelligence performs computations, courts distinguish computational processing from legal concept November 2025 revised guidance [10] expanded on these principles, providing additional examples and emphasizing claim-by-claim analysis.
TABLE II
COMPARATIVE
ANALYSIS
OF ARTIFICIAL INTELLIGENCE INVENTORSHIP AND OWNERSHIP ACROSS JURISDICTIONS Dimension United States United Kingdom EPO India Inventor Definition Natural person; conception is mental act by human (Thaler v. Vidal) Natural person; Patents Act presumes human inventors (UK Supreme Court) Natural person; EPC designates inventor as source of inventive thought (J 8/20) Likely natural person; true and first inventor implies human agency Statutory Basis
35 USC §101, §115;
case law on conception Patents Act 1977 §7, §13 EPC Art. 60(1), Rule 19; Guidelines Patents Act 1970 §6, §2(1)(y) AI as Inventor? No. Requires Congressional action No. Requires Parliamentary amendment No. Requires EPC revision by States No explicit guidance. Likely no Ownership Route Inventor or assignee. Gap if no human inventor Inventor or successor. Ownership of AI insufficient Inventor or successor. Gap if AI invents True and first inventor or assignee. §6(b) employer ownership USPTO 2024 & 2025: human must contribute significantly to conception AI-Assisted Guidance No formal guidance. Tool theory implicit No AI-specific guidance. Natural person rule No formal guidance Disclosure Inventor oath (§115). False statements risk invalidity Statement of inventor (§13) Designation (Rule 19) Forms require names. False grounds for rejection Policy Stance Courts: for Congress.
USPTO: guidance within existing law Courts: for Parliament. Status quo Board: interpretation limited. Reform needs consensus No formal position Impact Document human contributions. Litigation risk Uncertainty for AI R&D Harmonization limits flexibility Uncertainty. No clear path for AI-generated of conception implying consciousness, intent, and understanding. deemed inventor. If artificial intelligence autonomously generates invention with minimal employee contribution, employer ownership may not attach.
3 Incentive Theory
Patent law incentivizes innovation by granting exclusive rights to inventors in exchange for disclosure. Artificial intelligence lacks motivation, cannot be incentivized by property rights, and does not respond to legal incentives. Recognizing artificial intelligence as inventor would sever link between patent grant and incentive.
3 Deeming Provisions
Some scholars propose deeming provisions where person who owns, controls, or commissions artificial intelligence could be deemed inventor by statute, filling ownership gaps. However, this creates legal fiction. 4) Sui Generis Rights: Alternative is creating distinct protection category for artificial intelligence-generated inventions separate from traditional patents. Rights could vest in artificial intelligence owner or developer without requiring inventor designation.
4 Legal Personhood
Inventors must be capable of holding rights and entering legal relationships. Artificial intelligence lacks legal personhood in all studied jurisdictions. It cannot own property, contract, sue, or be sued. Without personhood, artificial intelligence cannot be inventor in legal sense. C. Practical Implications
1 Filing Strategy
Applicants using artificial intelligence face disclosure dilemmas. Fully disclosing artificial intelligence role might undermine inventorship claims. Understating artificial intelligence contributions risks inequitable conduct or later invalidity challenges. USPTO guidance attempts to provide safe harbors: applicants should document how human inventors contributed to conception. B. Ownership Theories When artificial intelligence contributes significantly to invention, ownership questions become complex.
1 Assignment from Human Inventor
If human qualifies as inventor, standard assignment mechanisms apply. Human inventor can assign rights to employer, artificial intelligence developer, or other party. This approach maintains doctrinal consistency but depends on human making significant contributions.
2 Examination and Litigation Risk
Patent examiners must evaluate inventorship based on application disclosures. Artificial intelligence-assisted inventions may face heightened scrutiny. In litigation, defendants may challenge inventorship, seeking to invalidate patents by showing artificial intelligence, not named inventors, performed conception.
2 Employer Ownership
Employment doctrines can vest ownership in employer when employee invents using artificial intelligence in employment course. This works if employee is would vest in artificial intelligence owner or designated legal representative. Invention Created Evaluation: Could provide strong incentives by ensuring all artificial intelligence-generated inventions are protectable. Weak accountability as artificial intelligence cannot be held responsible. Enforcement would depend on holding artificial intelligence owner liable indirectly. Significant administrative burden requiring patent offices to establish artificial intelligence registration systems. Low harmonization potential as recognizing artificial intelligence personhood is profound legal change. Raises philosophical and ethical concerns about conferring rights on non-conscious entities. Did natural person make significant contribution?
No Yes Human is inventor. Standard ownership Policy Choice Option C: AI personhood Option B: Controller deemed inventor Option D: Sui Option A: No patent generis right D. Option D: Sui Generis Rights Fig. 3. Decision flowchart for inventorship and ownership determination under different policy options. Create new category of intellectual property protection specifically for artificial intelligence-generated inventions, distinct from patents. Rights could have shorter terms, different disclosure requirements, and vest automatically in artificial intelligence owner. VII. POLICY OPTIONS FRAMEWORK Fig. 3 presents decision flowchart for inventorship and ownership determination. Evaluation: Provides protection for artificial intelligencegenerated inventions without distorting traditional patent doctrine.
Moderate accountability as artificial intelligence owner holds sui generis right and can be held responsible for misuse. Requires new legislation and administrative infrastructure. However, simpler requirements might reduce examination burden. Difficult harmonization requiring international treaties to function effectively across borders. Primary risk is creating confusion and forum-shopping. A. Option A: Status Quo Maintain current doctrine: inventors must be natural persons, and artificial intelligence is characterized as tool. Artificial intelligence-assisted inventions are patentable if human made significant contributions to conception. Artificial intelligence-generated inventions without substantial human contribution are not patentable.
Evaluation: Preserves incentives for humans who use artificial intelligence as research tool but may reduce incentives to develop fully autonomous artificial intelligence inventors. Strong accountability as humans remain responsible. Relatively straightforward for patent offices though examiners face challenges evaluating artificial intelligence contributions. High harmonization potential as major jurisdictions already adopt this approach. Primary risk is leaving genuinely autonomous artificial intelligence-generated inventions unprotected. E. Criteria-Based Comparison Table III compares four options across key criteria. Based on this evaluation, Options A and B appear most viable. Option A requires only guidance refinement while Option B could be implemented through targeted statutory amendments. Options C and D face significant feasibility and coordination challenges. B.
Option B: Deeming Provisions
VIII. INDIA-FOCUSED
ANALYSIS
Enact statutory provisions deeming human who owns, controls, trains, or prompts artificial intelligence system to be inventor of artificial intelligence outputs, subject to defined conditions. India has not yet confronted artificial intelligence inventorship litigation, but proactive engagement is essential given India’s growing role in artificial intelligence research and development. Evaluation: Fills ownership gaps, ensuring artificial intelligence-generated inventions can be patented and incentivizing investment in artificial intelligence development. Moderate to strong accountability as deemed inventor bears legal responsibilities. However, deeming someone inventor who did not truly conceive creates legal fiction. Requires legislative amendment to define deeming conditions. Moderate harmonization potential as countries could adopt similar provisions through treaties. A.
Current Gaps in Indian Patent Law
1 Ambiguity on AI as Inventor
Patents Act, 1970 does not explicitly address whether inventors must be natural persons. While section 6 reference to true and first inventor likely presumes human inventors, absence of explicit statutory language or case law creates uncertainty.
2 Ownership Challenges
Section 6 allows patents to be granted to inventors or their assignees. If artificial intelligence generates invention autonomously, there is no human inventor from whom to derive title. Act does not provide deeming mechanism or sui generis route for such inventions. C. Option C: Limited Legal Personhood Grant artificial intelligence systems limited legal personhood, enabling them to be named as inventors. Ownership TABLE III EVALUATION OF POLICY OPTIONS FOR ARTIFICIAL INTELLIGENCE INVENTORSHIP AND OWNERSHIP Criterion Option A (Status Quo) Option B (Deeming) Option C (AI Personhood) Option D (Sui Generis) Innovation Incentives Moderate. Incentives for AI-assisted only High. Fills gaps High. All protectable Moderate to High Accountability Strong. Humans accountable Moderate. Fiction may weaken integrity Weak. Indirect via owner Moderate. Attenuated connection Administrability Moderate.
Requires guidance Moderate. Needs conditions Low. Infrastructure needed Low to Moderate Harmonization High. Jurisdictions converging Moderate. Treaty possible Low. Consensus unlikely Low. New treaties needed Doctrinal Coherence High. Maintains tradition Moderate. Legal fiction strains doctrine Low. Alters foundation Moderate. Adds complexity Abuse Risk Moderate. Overstate contributions Moderate to High. Strategic claims Moderate. Shell entities Moderate. Forumshopping
3 Disclosure and Ethical Concerns
Application forms require applicants to state inventor names. If applicant using artificial intelligence names human with minimal contribution, this could constitute false information, grounds for rejection or revocation. Conversely, if applicant attempts to name artificial intelligence or leave inventor field incomplete, application would be rejected. preventing evergreening. Policymakers must ensure reforms addressing artificial intelligence inventorship do not undermine these goals. Patent offices should rigorously apply patentability criteria regardless of artificial intelligence involvement.
2 International Coordination
India is signatory to TRIPS Agreement and engages in WIPO forums. Any Indian reforms should consider international trends to facilitate cross-border filing and enforcement. Aligning with USPTO-style guidance promotes harmonization without requiring treaty amendments. B. Comparative Lessons for India
1 Adopt USPTO-Style Guidance
India should issue administrative guidance clarifying inventorship for artificial intelligence-assisted inventions, modeled on USPTO guidance. Guidance should affirm inventors must be natural persons, clarify that use of artificial intelligence does not automatically disqualify person from inventorship, provide factors for evaluating significant human contribution to conception, and offer examples illustrating scenarios. 3) Capacity Building: Examiners will need training to evaluate artificial intelligence-assisted inventions. India’s Patent Office should invest in examiner education on artificial intelligence technologies, conception analysis, and disclosure review. D. Proposed Roadmap
1 Short-term (1-2 years)
Issue administrative guidance on artificial intelligence-assisted inventions, clarifying inventorship standards and providing examples
2) Medium-term
(3-5 years): Introduce legislative amendment adding deeming provisions for artificial intelligence-generated inventions. Enhance disclosure requirements in application forms 3) Long-term (5+ years): Evaluate effectiveness of deeming provisions and consider whether sui generis rights or other mechanisms are needed. Engage in international coordination efforts through WIPO
4 Ongoing
Conduct training for examiners, establish stakeholder consultation processes, and monitor case law developments globally This guidance could be issued by Controller General of Patents through official notice or manual revision.
2 Consider Legislative Amendment
In medium term, India could enact deeming provision in Patents Act. For example, new sub-section under section 6 could state: Where invention is generated by artificial intelligence system, natural person who directed, trained, or controlled system for purpose of solving specific technical problem may be deemed inventor, provided that person made material contribution to development or deployment of system. This provision would fill ownership gaps while maintaining human accountability. 3) Enhance Disclosure Requirements: India should consider requiring patent applicants to disclose use of artificial intelligence in inventive process. This could be implemented through checkbox or statement in application forms. Enhanced disclosure would assist examiners in evaluating inventorship and promote transparency. IX.
CONCLUSION
AND FUTURE RESEARCH This research examined collision between artificial intelligence and patent law through comparative analysis of DABUS litigation and related guidance in US, UK, EPO, and implications for India. Findings reveal global consensus that inventors must be natural persons, grounded in statutory language, doctrinal coherence, and policy rationales including incentive C. Policy Considerations
1 Balancing Innovation and Access
India’s patent policy historically emphasizes access to essential goods and theory and accountability. However, this consensus leaves unresolved question of how to address artificial intelligencegenerated inventions where human contribution is minimal or absent. trine. DABUS litigation has clarified that under current law, artificial intelligence cannot be inventor. Yet this clarity leaves deeper questions unresolved: how should legal systems allocate ownership of artificial intelligence-generated inventions, incentivize innovation, and maintain accountability? An ownership gap exists: without human inventor, standard assignment mechanisms fail, and such inventions may go unprotected. This gap creates perverse incentives to characterize artificial intelligence contributions dishonestly or to withhold artificial intelligence-generated inventions as trade secrets.
Different jurisdictions may adopt different approaches based on their legal traditions, economic priorities, and policy values. What is essential is that reforms be made thoughtfully, with attention to doctrinal coherence, practical administrability, and international coordination. Framework and analysis presented aim to support policymakers, patent offices, and stakeholders in navigating these complex choices. Four policy options are available: maintaining status quo with refined guidance, enacting deeming provisions, granting artificial intelligence limited legal personhood, or creating sui generis rights. Evaluation suggests Options A and B are most viable. Option A aligns with current international trends and requires minimal legislative change. Option B fills ownership gaps while preserving human accountability, though it introduces legal fiction.
As artificial intelligence capabilities continue to advance, patent law must evolve—not by abandoning foundational principles, but by adapting them to new technological realities. Challenge is to preserve patent system core functions—rewarding innovation and promoting disclosure—while accommodating future in which invention is no longer exclusively human endeavor. For India, we recommend issuing administrative guidance in near term, followed by consideration of deeming provisions through legislative amendment. Enhanced disclosure requirements and examiner training will support consistent application. India’s approach should balance promoting artificial intelligence innovation with safeguarding access to essential inventions and preventing strategic manipulation.
REFERENCES
[1] World Intellectual Property Organization, “WIPO Technology Trends 2019: Artificial Intelligence,” Geneva, Switzerland, 2019. [Online]. Available: https://www.wipo.int/publications/en/details.jsp?id=4386 [2] World Intellectual Property Organization, “Artificial Intelligence and Inventorship,” WIPO Meeting Document, Sept. 2023. [3] World Intellectual Property Organization, “AI Inventions: Policy Con- A. Future Research Directions siderations,” WIPO Publication, 2023. [4] European Patent Office Legal Board of Appeal, “Decision J 8/20 of Several areas warrant further investigation:
1 Empirical Studies
Empirical research is needed on prevalence of artificial intelligence-assisted and artificial intelligence-generated inventions in patent filings. Surveys of inventors and patent attorneys could reveal how artificial intelligence is currently used and how applicants navigate disclosure dilemmas. Analysis of patent data could identify trends in artificial intelligence-related filings.
21 December 2021
Designation of Inventor/DABUS,” Official Journal EPO, 2022.
[5] European Patent Office, “Press Communique ́: Decision J 8/20,” July 2022. [6] European Patent Office, “Guidelines for Examination: Designation of the Inventor,” Part A, Chapter IV, 2023. [7] Thaler v. Vidal, No. 21-2347 (Fed. Cir. Aug. 5, 2022). [8] Thaler v. Comptroller-General of Patents, [2023] UKSC 49 (UK Supreme Court, Dec. 20, 2023). [9] United States Patent and Trademark Office, “Inventorship Guidance for
2 Economic Modeling
Economic models could assess incentive effects of different policy options. Would deeming provisions increase artificial intelligence research and development investment? Would lack of protection for artificial intelligence-generated inventions drive such inventions into secrecy? AI-Assisted Inventions,” 89 Fed. Reg. 10043, Feb. 13, 2024. [10] United States Patent and Trademark Office, “Revised Inventorship Guidance for AI-Assisted Inventions,” 90 Fed. Reg. 78234, Nov. 28, 2025. [11] Government of India, “The Patents Act, 1970,” Act No. 39 of 1970, as amended. [12] R. Plotkin, “What Does It Mean to Be an Inventor? The Conception
3) Comparative Analysis
This research focused on US, UK, EPO, and India. Future work could examine artificial intelligence inventorship approaches in other major patent jurisdictions such as Japan, China, and South Korea. Requirement in US Patent Law,” Intellectual Property Today, vol. 16, no. 9, pp. 22–25, Sept. 2009. [13] L. Vertinsky and T. M. Rice, “Thinking About Thinking Machines: Implications of Machine Inventors for Patent Law,” Boston University Journal of Science & Technology Law, vol. 8, pp. 574–613, 2019. [14] B. Hattenbach and J. Glucoft, “Patents in an Era of Infinite Monkeys 4) Intersection with Other IP Rights: Artificial intelligencegenerated works raise similar issues for copyright (authorship) and design rights. Comparative analysis across intellectual property regimes could identify common principles and divergent approaches.
and Artificial Intelligence,” Stanford Technology Law Review, vol. 19, pp. 32–51, 2015. [15] S. Yanisky-Ravid and X. Liu, “When Artificial Intelligence Systems Produce Inventions: The 3A Era and an Alternative Model for Patent Law,” Cardozo Law Review, vol. 39, pp. 2215–2263, 2018. [16] R. Abbott, “I Think, Therefore I Invent: Creative Computers and the 5) Ethical and Philosophical Dimensions: Deeper philosophical analysis of artificial intelligence creativity, consciousness, and moral status could inform debates about artificial intelligence personhood. Ethical analysis of artificial intelligence role in knowledge production would enrich policy discussions. Future of Patent Law,” Boston College Law Review, vol. 57, no. 4, pp. 1079–1126, 2016. [17] B. Fraser, “Computers as Inventors: Legal and Policy Implications of Artificial Intelligence on Patent Law,” SCRIPTed, vol. 13, no. 3, pp.
305–333, 2016. [18] L. Schuster, “Artificial Intelligence and Patent Ownership,” WIPO Magazine, Aug. 2018. [19] D. Gervais, “The Machine as Author,” Iowa Law Review, vol. 105, pp. B. Concluding Remarks 2053–2106, 2020. [20] A. Bridy, “Coding Creativity: Copyright and the Artificially Intelligent Patent law stands at inflection point. Artificial intelligence challenges assumptions embedded in centuries of legal doc- Author,” Stanford Technology Law Review, vol. 5, 2012.