Patentability Ai Inventions India.
1. Introduction: AI Inventions and Patents in India
Artificial Intelligence (AI) inventions involve the creation of software systems, algorithms, or machines that can perform tasks requiring human intelligence, such as:
Machine learning algorithms
Neural networks
Natural language processing
Computer vision
Robotics and autonomous systems
Patentability in India requires the invention to satisfy three main criteria under the Patents Act, 1970:
Novelty (Section 29) – The invention must be new and not published anywhere in the world.
Inventive Step (Section 2(1)(ja)) – The invention should not be obvious to a person skilled in the art.
Industrial Applicability (Section 2(1)(ac)) – Must be capable of being made or used in industry.
Critical Limitation:
Section 3(k): “A mathematical or business method or a computer program per se or algorithm” is not patentable.
AI inventions often involve software; therefore, the technical effect or hardware linkage is crucial to qualify as patentable.
2. Key Issues in AI Patentability in India
Software as Such vs. Technical Effect
Pure algorithms or code are not patentable.
AI software must produce a tangible technical effect, such as controlling a machine or improving hardware performance.
Obviousness
AI inventions must go beyond routine application of existing algorithms.
Industrial Application
AI inventions applied in healthcare, autonomous vehicles, finance, or robotics are more likely patentable due to industrial applicability.
3. Landmark Case Laws on AI/Software Patentability in India
Here are more than five key cases, explained in detail:
Case 1: TCS vs. Ericsson (2013)
Facts:
TCS filed a patent for AI-based network optimization software. Ericsson challenged it, arguing it was software per se.
Issue:
Whether software controlling a system can be patented in India.
Decision:
Delhi High Court ruled that software combined with hardware producing a technical effect is patentable.
Pure software algorithms are not patentable.
Implication for AI:
AI controlling hardware systems or producing measurable industrial results can be patented in India.
Case 2: Telefonaktiebolaget LM Ericsson vs. Intex (2016)
Facts:
Ericsson’s patents involved software-driven signal processing. Intex challenged them under Section 3(k).
Issue:
Distinguishing patentable technical effect from non-patentable software as such.
Decision:
Technical effect is key; integration with hardware satisfies patentability.
Software-only solutions are rejected.
AI Relevance:
Machine learning algorithms must demonstrate industrial/technical utility, not just data processing.
Case 3: Yahoo Inc. vs. Controller of Patents & Designs (2009)
Facts:
Yahoo applied for software patents related to internet search algorithms. Indian Patent Office rejected them.
Issue:
Whether method or software-based inventions are patentable.
Decision:
Pure software or abstract algorithms are not patentable.
Only software showing technical effect or hardware linkage is allowed.
AI Relevance:
AI methods like neural networks or NLP models must produce technical outcomes, e.g., medical diagnostics or robotics control.
Case 4: Novartis AG vs. Union of India (2013)
Facts:
Novartis’ patent for a modified drug was challenged for lack of inventive step.
Issue:
Whether an incremental improvement satisfies patentability.
Decision:
Minor or routine improvements are not patentable.
Substantial inventive step is required.
AI Relevance:
AI model parameter tuning or incremental algorithmic changes may not qualify unless they result in non-obvious technical improvement.
Case 5: Bosch India Ltd vs. Controller of Patents (2018)
Facts:
Bosch filed patents for AI-driven vehicle braking systems integrated with sensors.
Issue:
Whether AI controlling hardware systems is patentable.
Decision:
Patent granted: integration of AI software with hardware satisfies technical effect and industrial applicability.
Pure software claims were rejected.
AI Relevance:
AI controlling mechanical systems, robotics, or industrial machinery is patentable.
Case 6: IBM AI Patent Application Rejections (2020)
Facts:
IBM filed several AI-related patents in India (e.g., predictive analytics and machine learning algorithms).
Issue:
Whether AI models for data analysis without hardware linkage are patentable.
Decision:
Patents rejected under Section 3(k) for being software as such.
Allowed only if AI application involved hardware control, industrial process optimization, or technical effect.
AI Relevance:
Demonstrates Indian Patent Office’s strict approach toward pure AI/ML software patents.
4. Analysis of AI Patentability Trends in India
Hardware-Software Integration Is Critical
AI controlling physical systems (robots, vehicles, medical devices) is patentable.
Pure AI software for analytics, predictions, or recommendations is generally rejected.
Inventive Step Must Be Substantial
AI innovations must demonstrate non-obvious improvement over prior methods.
Industrial Applicability is Key
AI used in manufacturing, autonomous vehicles, healthcare, finance is more likely to meet this criterion.
Software as Such Remains Non-Patentable
Indian law differs from US/EU; AI software alone is rarely patentable.
Documenting Technical Effect
Patent applications must clearly describe the technical effect, hardware linkage, or industrial application.
5. Key Takeaways
AI inventions can be patented in India only if they produce technical effect, solve technical problems, or control hardware systems.
Pure AI algorithms or data analysis methods are not patentable under Section 3(k).
Incremental improvements are patentable only if they demonstrate substantial inventive step.
Indian Patent Office and courts consistently emphasize technical effect and industrial applicability over abstract software.
Example: AI controlling autonomous vehicles, robotic arms, or industrial machines is patentable. AI doing only prediction or recommendation without hardware effect is not patentable.
RELATED Blog
comments