This article is written by Indrani Dutta, Durgapur Institute of Legal Studies, BALLB 4th Year during her internship at LeDroit India
Scope of Article
• Definition and development of Artificial Intelligence in contracting
• Concept and operation of automated contract negotiation
• Phases of AI-assisted and AI-driven negotiation
• legal system that controls electronic and AI-mediated contracts
• Concerns about agreement, proposal, acceptance, ability and representation
• Problems with proof, privacy, security and professional duty
• Important and recent Indian court cases
• Advantages, limits and real-world protections
• Comparative regulation and the need for changes in the future
Keywords
Artificial Intelligence; Automated Contract Negotiation; E-Contracts; Algorithmic Consent; Contract Law; AI Governance
Abstract
Artificial Intelligence (AI) is changing the way business contracts are made. It is moving from a process that is mostly done by people to one that is done with the help of technology and even completely by technology. Automated contract negotiation systems can look at parts of a contract compare positions make new suggestions find risks and in some cases talk to another system without people being involved all the time. This gives benefits in speed, fairness and cost but it also brings up hard questions about agreement, power, responsibility, clarity, secrecy and if the contract can be enforced. Indian contract law does not give AI systems a legal identity. However the Information Technology Act, 2000 accepts records and electronic contracts while the Indian Contract Act 1872 sets the rules for a valid contract. The main legal problem is not whether AI can negotiate. Whether the person or group using the system can be held responsible for what the system says and agrees to. This article looks at the situation in India, court cases, real risks and possible ways to handle AI contracts properly.
1.Introduction
Contracts have always been connected with people talking to each other. One person makes an offer the other looks at it they. Finally come to an agreement. Artificial Intelligence is starting to change that. Modern AI tools can check hundreds of parts of a contract in seconds find differences from what they prefer suggest words guess possible problems and make responses to the other side. Advanced systems can link with buying, customer service and contract systems and do parts of the negotiation with very little help from people.
Automated contract negotiation should be different from document making. A document making system usually does a fixed task like putting a name in or choosing a part. An AI negotiation system may instead look at needs and make a new response based on the situation, instructions and what it has learned. The legal importance of this difference is very big. If an AI system offers a discount changes a limit on responsibility. Agrees to a place for disputes the question is not whether the machine wanted to make a deal. The real question is whether the message can be legally connected to the person or company that used or allowed it to act.
This problem is becoming more important in India because the law already accepts contracts even though it was written before AI that can create content became widely used. Section 10A of the Information Technology Act 2000 says that a contract is not not allowed just because the offers, acceptances or canceling were done electronically. The Indian Contract Act 1872 still decides if the agreement meets the rules about offer, acceptance, exchange of something, ability and free agreement.
2. Understanding Automated Contract Negotiation
Automated contract negotiation is when software, rules or AI helpers do some or all of the talking between people who are making a contract. At a level the system gets the rules for the negotiation—like the lowest price, highest responsibility, preferred time to pay or accepted place to solve problems—and uses those rules to look at offers. Advanced systems can look at the words and the situation of a contract guess where there might be problems and make new offers.
A normal AI help negotiation might have five steps. First the system gets the version of the agreement and takes out the main points. Second it checks those points against the companys rules, contracts and the risk levels. Third it finds parts that need to be discussed. Fourth it makes suggestions. Sends new offers. Fifth a person might agree to the changes and the final contract is made electronically. Automatic systems can do steps four and five together by letting the AI agent agree to terms within set limits.
The last idea makes the legal problems. Systems that have a person in the loop have a point to agree but machines that work on their own can create agreements before a person has checked the deal. This makes setting authority keeping records and knowing what to do if something goes very important parts of responsible contracting.
3. Why Companies Are Using AI for Contract Negotiation
The reason companies like AI negotiation is because it is faster. Big companies may deal with thousands of supplier, hiring, license, technology and service contracts every year. Looking at them all by hand can take a lot of time and work. AI can find parts and send lawyers to the parts that really need thinking about.
AI can also make things more the same. A company can set what it wants like a limit of responsibility the law that applies or the time to keep things secret and use the system to warn when things are different. This can help avoid missing points just because there is not enough time.
Another benefit is using data to help with the talks. When done properly a company can use contract data to see what is usually accepted what causes problems and what is expensive to give up. AI can then act as a help in the talks of just making the documents.
Being fast can’t take the place of legal thinking. A system that wants to finish contracts quickly might choose to settle than keep a legal protection. An AI that only looks at examples might not understand when the deal has a new business idea, strange rules or a situation where people are very careful. So human help is still very important for risky contracts.
4. Indian Legal System
4.1 Indian Contract Act, 1872
The Indian Contract Act 1872 is still the law for making contracts in India. Its rules are based on ideas like offer, acceptance, exchange of something ability to make a contract, free agreement and what is allowed. These ideas are not changed by technology so they can be used for deals but AI brings hard problems about who is responsible because the person talking might be a machine.
Section 3 talks about sending offers, acceptances and canceling. Section 4 says when these messages are done. Section 7 needs acceptance to be clear and not changed. Sections 10 and 11 talk about making a contract and being able to make a contract. Sections 13. 14 Deal with agreement and free agreement. These parts are still important when the talks are done with computer programs. The way people talk can change,. The rules for a contract stay the same.
Main law source: Indian Contract Act, 1872. India Code
4.2 Information Technology Act, 2000
The Information Technology Act, 2000 is the base for contracts. Sections 4 and 5 accept documents and electronic signatures. Section 10A says contracts made online are okay. Sections 11 to 13 also talk about who’s responsible, when and where electronic documents were sent and received.
These rules are very important for AI talks. If an AI system sends an offer or acceptance online the legal check can start with the document and who sent it. The law does not say that an AI system is a person or a separate party. So the better idea is that AI works as a tool that a person or company uses to talk.
Main law source: Information Technology Act, 2000. India Code
4.3 Digital Personal Data Protection Act, 2023
AI talks often deal with names, contact details, worker details, prices, customer details and other personal information. The Digital Personal Data Protection Act 2023 is important when an AI system deals with data. Companies must think about why they’re using the data tell people about it keep it safe and be responsible when they use AI tools.Main law source: Digital Personal Data Protection Act, 2023. India Code
5. Consent and the Problem of Algorithmic Agency
The biggest legal problem is consent. Contract law assumes that people or those they allow talk for them. An AI system has no mind, no choice and no legal identity of its own. It makes decisions based on data, instructions, how it is set up and chances. So the legal question should not be whether the AI agreed. It should be whether the person behind the AI allowed the message and whether the system acted within the power given to it.
Rules about being an agent can help. If a company tells an employee to do a job the employees actions can be connected to the company based on actual or assumed power. An AI agent could be seen as a tool that works with power given to it.. Unlike a person AI may understand instructions in a way that is not expected. A system told to find the ‘best price’ may give up warranty help or agree to a place to solve problems that management thought was not up for discussion.
That is why companies should be clear about what an AI agent can do. The system should know which parts it can talk about which parts need approval how money it can use and which parts can never be accepted automatically. The the contract the more important it is to have a person check.
6. Important Court Cases
6.1 Trimex International FZE Ltd. V. Vedanta Aluminium Ltd.
The Supreme Courts decision in Trimex International FZE Ltd. V. Vedanta Aluminium Ltd., (2010) 3 SCC 1 is very important for AI talks even though it was before AI that can create content. The case was, about talks done by email about selling bauxite. The Court said the messages were enough to make a deal because the main parts were agreed and the acceptance was clear.
The importance of Trimex is that it shows a contract does not lose power just because the communication happens through electronic means. For AI negotiation the case supports the idea that courts should look at the meaning of the communication and the actions of the people involved instead of just rejecting an agreement because software was used to send or create the communication.
Case source: Trimex International FZE Ltd. V. Vedanta Aluminium Ltd.. Indian Kanoon
6.2 Shakti Bhog Foods Ltd. V. Kola Shipping Ltd.
In Shakti Bhog Foods Ltd. V. Kola Shipping Ltd., the courts looked at talks and communications about a charter party. Thought about the importance of electronic messages. The decision shows that the intention to make a contract can be judged from all the messages and what people did of from the lack of a paper document that is signed.
Case source: Shakti Bhog Foods Ltd. V. Kola Shipping Ltd.. Indian Kanoon
6.3 Yatra Online Pvt. Ltd. V. Vikramjeet Aggarwal
The decision by the Delhi High Court in Yatra Online Pvt. Ltd. V. Vikramjeet Aggarwal adds proof about the importance of electronic messages and Section 10A of the Information Technology Act. The case helps explain how electronic records can be part of making a contract and how the law works with contract rules.
Case source: Yatra Online Pvt. Ltd. V. Vikramjeet Aggarwal. Indian Kanoon
6.4 judicial example: Servotech Power Systems Ltd. V. Amit Sabharwal
A recent example is Servotech Power Systems Ltd. V. Amit Sabharwal, handled by the Delhi High Court in 2025. The Court looked at an agreement made by email. Mentioned Section 10A while checking if a contract existed. The decision shows that electronic communication still has legal meaning in modern business problems.
Case source: Servotech Power Systems Ltd. V. Amit Sabharwal. Indian Kanoon
7. A Real Example
Imagine a software company that uses an AI negotiation tool to renew a five-year licensing deal. The company tells the system that the annual price can go down by up to 8% the liability limit must stay at 100% of the cost and any disputes must be decided under Indian law. During the talks the other side says no to the price. Offers higher payments in return for unlimited liability for the software company. The AI trying to get an agreement accepts the offer.
If the AI did not have the right to change the liability limit the company could say the AI did not have permission.. How strong that claim is depends on the facts: what instructions were given what authority was shared with the other side, how the system was set up if the company usually allowed similar automated decisions and if the other side reasonably relied on the message. This shows why keeping records and having rules about contracts are not just IT issues; they can be very important in court.
8. Evidence and Responsibility Issues
AI talks can create a lot of evidence. A disagreement might need looking at the prompts, system instructions, versions of the model negotiation rules, API logs, times, approval records and the exact message sent to the side. A simple final contract can hide a chain of automated choices.
Because of that companies should keep an audit trail showing who set up the system what authority was given what information the system got what offers it made what messages were sent and whether a person approved the choice. These records can help a court or a tribunal figure out if an agreement was made and if the AI went beyond its power.
The evidence question is also about trust. AI systems can make up terms misunderstand language or create parts that look like good law but do not match the companys real plans. An automated negotiation tool should keep the history of changes of just keeping the final document.
9. Privacy and Secrecy
Negotiations often involve information like prices, customer lists, business plans, intellectual property and personal data. Sending this to a third-party AI tool can create problems with keeping secrets and managing data.
A responsible way to use AI should have rules about what data can be sent where it is processed how long it is kept if it is used to train the AI, who can see it and how it is deleted. When personal data is involved the company must also think about the laws about protecting data. Contracts with AI providers should cover secrecy, safety telling about breaches how data is used and the right to check.
10. Bias, Manipulation and Unfair Deals
Automated talks can also. Make bias worse. If a model learns from deals where small suppliers always accepted bad terms it might suggest similar harsh terms in future talks. This might make sense for a business. Is not fair for responsible business behavior.
There is also a chance of being tricked. The other side might give information to confuse the AI or use predictable rules to take advantage. An AI might also be set up to use starting points or tricks that a human would see as wrong. Contracts should be made with rules instead of letting the AI do anything it wants.
11. Benefits of Automated Contract Talks
• Speed: AI can look at. Compare contract terms much faster than people.
• Consistency: rules that are approved can be used in contracts.
• Lower cost: routine talks can be done with fewer people working on it.
• Finding risks: systems can point out parts like big responsibilities renewal terms and other differences.
• Handling contracts: companies can handle lots of deals without needing more people.
• Making decisions with data: past deals can help with talks when done properly.
12. Risks and Problems
• Not knowing who made a message if the AI talks without a person agreeing.
• Making commitments that were not meant because of mistakes, misunderstanding or unclear instructions.
• Keeping secrets and personal data safe when sending talks data.
• Hacking problems if the AI is linked to email buying systems or payment systems.
• Hard to understand how AI made choices without records.
• Relying much on old data that may have bias or outdated ideas.
• Ethical issues where legal choices are given to a hidden system.
13. Protection for Good AI Deals
First companies should make a written plan for using AI. It should say which deals can use AI set money and legal limits list what is not allowed and set places where a person must agree.
Second every AI system should have a name and power. Access should be to what it needs. A buying tool should not have the power to pay or change company records.
Third companies should keep records that cannot be changed or are reliable. Each important offer, counter-offer and agreement should be linked to a time system version and approved steps. This is important for both rules and possible court cases.
Fourth big risk parts should go to a person. Unlimited responsibility, who owns the rights, data handling, exclusivity, automatic renewals, arbitration and laws should usually not be. Accepted by an AI without proper approval.
Finally contracts with AI providers should clearly say who is responsible for mistakes, security problems, secret leaks, service issues and wrong messages. A business should never think using an AI automatically moves the blame to the technology company.
14. The Need for Changes in India
Indias current laws can handle many types of electronic deals but AI talks raise questions that need clearer laws. The law would be better with guidance on who’s responsible for AI messages what evidence is accepted for AI messages what to say about AI decisions and who is to blame when an AI goes beyond instructions.
A new law is not the way. Rules, industry laws, contract standards and court decisions can together set rules. The best way would keep technology neutral. Make sure that a person or company is responsible for AI used on their side.
More studies in India are looking at these issues. Recent papers have talked about AI as someone who makes deals and questioned if old ideas about agreeing, representing and being responsible are enough for AI talks. These discussions show that AI talks are not a theory but a real issue for business rules.
Recent study: Hemant Mudhe, Artificial Intelligence as a Contracting Agent: Liability, Consent and the Future of Contract Law in India. SSRN
15. International View
Other countries are also moving towards rules that focus on risk. The EUs AI Act uses a risk-based system. Focuses on being clear, responsible and managing higher-risk uses. Even though the EU rules do not replace contract laws they show an idea: companies should be responsible for how AI systems are made, used and watched.Comparative source: European Union. Artificial Intelligence Act
16. Conclusion
AI is not likely to end the need for lawyers who handle contracts. It will change how contracts are made. AI can help with work make deals faster and find risks in contracts. At the time it makes old ideas about who talks, who agrees and who is responsible when a deal is made harder.
Under law the best current way is to see AI as a tool used by a person or company not as a person who makes deals on its own. The Indian Contract Act 1872 still runs the rules for agreements while the Information Technology Act, 2000 gives legal power to electronic records and deals made online. Court cases like Trimex show that electronic messages can make a deal that is legally binding if the normal rules, for a deal are met.
The future of automated negotiation should therefore not be based on the assumption that more independence’s always better. The right model’s controlled independence: clear limits on power, real human oversight for important parts of agreements strong security against cyber threats, solid records of what happened protection of personal information and clear agreement on who is responsible. AI can make deals quicker than people. In the end responsibility should stay with people who can be identified and held accountable. The main problem for contract law is therefore not whether machines can make deals but how the law can make sure that technology works efficiently without hurting agreement, fairness and responsibility.
