Introduction
This guide is aimed at global in-house counsel, private practice lawyers and risk and compliance professionals. It can be used in most jurisdictions and has a specific focus on the United States. It will assist in understanding how artificial intelligence (AI) is being used alongside smart contracts to automate and streamline the contract management process, the common risks associated with this, and how the face of contractual liability and enforceability considerations are changing against this backdrop.
This guide covers:
- Evolution of AI in contract automation
- The application of AI and smart contracts
- Commercial uses in practice
- Legal status and liability
- Risks of enforcement action
This guide can be used in conjunction with the following How-to guide: Understanding the risk of negligence claims when using AI.
Section 1 – Evolution of AI in contract automation
AI involves the use of sophisticated technology to mimic the actions or thought processes of humans in the performance of tasks such as decision-making and problem-solving. It is becoming an increasing presence in all aspects of daily life and its use is transformative across many industry sectors, including commercial contracts. Businesses are increasingly using AI technology to improve efficiencies and workstream productivity, and generate increased revenue through enhanced contract analysis and by streamlining contract negotiation and drafting processes.
Artificial intelligence (AI) may be defined broadly as using machines to simulate human intelligence to undertake ‘human’ tasks and functions (eg, problem-solving and reasoning). The researchers Darrell M. West and John R. Allen have identified three essential qualities of AI:
- intentionality, or the ability to make decisions, often using real-time data;
- intelligence, or using data to look for underlying trends; and
- adaptability, meaning the ability to learn and adapt as decisions are made.
Experts in the field, including many who have worked on the development of AI, have raised warnings about its use, saying that AI poses an existential threat to humanity on a par with nuclear war or pandemics. While it is certain that the rollout of AI-enabled business practices has the potential to transform and is making an impact across the global economy, it is important to be mindful of the liability risk, ensuring that there are sufficient safeguards in place to minimize the risk of harm, and to maintain ethical and responsible business practices as technology evolves.
Section 2 – The application of AI and smart contracts
2.1 What are smart contracts?
Smart contracts have been in use since the 1990s and are self-executing contracts that automate actions between the contracting parties. They are called ‘smart’ and ‘self-executing’ as the terms of the contract do not contain legal language but rather are written in computer code allowing contract execution to be fully automated without third-party intervention. Actions that would otherwise be completed and tracked by individual parties to the agreement are automated using technology.
There is no need for further negotiation of contractual terms, as a smart contract performs a party’s obligations automatically, when a given set of preconditions is met. This has the potential to greatly improve efficiency in the contract management process and reduces the risks associated with human error or fraud.
2.2 How do smart contracts work?
To establish the terms, the parties must agree how the transaction and data should look, the different scenarios that will trigger contract execution (ie, the ‘if and when . . . then’) conditions, and explore whether exceptions apply or what should happen in the event of a dispute. This enables a computer program to monitor and complete transactions and manage the contract accurately and as intended. When the pre-determined conditions are satisfied, the computer automatically performs certain actions (eg, processes payments, releases funds, or generates ticket sales).
Since smart contracts run automatically, the need for performance management and supervisory oversight by third parties is reduced. This saves time and creates efficiencies by cutting out paperwork and manual verification. For example, a smart contract might be set up so that payment is triggered upon the purchase and download of software. As soon as the software is downloaded, the program will automatically make payment. There is no need for human intervention, or direction to execute a command. This automation in payments helps to speed up the processing costs and reduces the prospect of making late or delayed payments because of human error. Prompts to pay are programmed into the system, and as soon as ‘trigger’ events are reached (eg, delivery of a product), the payment is made automatically.
2.3 Blockchain and smart contracts
Smart contracts often run using blockchain technology to automate the execution of contract terms.
Blockchain is a shared digital ledger and database that records transactions and enables parties to save and share data securely and transparently. Any time a smart contract is created, it is stored as a data block on the blockchain. A blockchain network can track orders, payments, accounts, production, and much more. In implementing the procedures to create a smart contract, the parties need to agree how the blockchain will represent transactions and data. Developers will then construct the contract based on the specifications, and parties will use either an in-house software developer or can outsource the project to a third-party IT provider.
The process is secure as it uses encryption techniques to ensure that the information cannot be altered. With this technology, each block of data is transferred to, and stored on, the blockchain. When certain trigger conditions are satisfied, the smart contract executes automatically. The smart contract’s execution is immediately broadcast to the blockchain. Once the transaction is complete, the smart contract and the blockchain update to reflect the completed transaction. Security is assured because once transactions are validated, they cannot be modified or erased, and the record of the smart contract is available for review at any time.
For additional information, see Quick view: Introduction to cryptocurrency and how it works (USA).
2.4 The integration of AI and smart contracts
The continuing evolution of blockchain technology and smart contracts shows the potential of AIto bring more intelligent, secure, and streamlined processing to blockchain-based smart contracts.
AI is a broad term. What is usually thought of as AI includes the creation of intelligent machines, usually based on or comprising computer systems, that simulate or mimic human learning, reasoning, and problem-solving. The broad term ‘AI’ encompasses the sub-fields of machine learning (ML) and deep learning (DL). ML is directed towards building systems that can learn from data, make decisions, and get ‘smarter’ over time. DL techniques enhance learning by absorbing huge amounts of data such as text, images, or video. AI is predictive and designed to react to information in a way that would mimic a human reaction to the same information.
This can be useful when contracting as AI can identify trends, make predictions, and identify risks that could impact a contract’s outcome. The vast amount of data that AI can process means that it may well identify trends and distinctions that the human mind does not have the capacity to process – arguably, AI has the potential to make smart contracts smarter.
2.4.1 Benefits of using AI in smart contract development
Some of the benefits of using AI-driven smart contracts include:
- Streamlined contract processing – AI-based methods may help inform contract negotiations by customizing contract terms during the negotiation and drafting process, or by examining contract language for potential ambiguities and inconsistencies. This will help to streamline transaction processing in contract negotiations and identify anomalies, leading to the creation of a tailored contract which saves on the time and costs associated with protracted contract negotiations.
- Contract automation – by automating tasks and decision-making, AI can help mitigate some of the risks associated with human errors.
- Predictive contract management – AI can be used to review smart contracts in real-time to detect unusual patterns, and AI algorithms can be trained to update contract terms based on real-time market conditions or to predict and prioritize transaction processing (eg, recommend expired products to recall, or execute transactions such as payments or select the most suitable shipping method). This helps developers and users to make faster decisions and more informed choices assessed against wider fact patterns.
- Enhanced security – by detecting and preventing fraudulent activity. ML algorithms can detect suspicious activity through analysis of transaction data. This proactive security approach expedites identification of a system's vulnerabilities and risks. This approach benefits all stakeholders and strengthens the trust in the network.
- Dynamic compliance and regulation adherence – AI can continuously and efficiently monitor changes in laws and regulations. Smart contracts will adapt and automatically redraft as needed to comply with new legal requirements. This reduces both parties' risk of non-compliance and potential legal issues, ensuring that contracts remain valid and enforceable over time.
- Improved data analysis and insights – AI system analyze vast amounts of data. Analyzing contract-related data will provide valuable insights and trends, and provide them quickly, as they develop. This will help organizations optimize their contracting strategies, identify opportunities for cost savings, and enhance business decision-making by taking advantage of data-driven insights.
2.4.2 Enhanced efficiency in complex transactions
AI can also fill in gaps in blockchain technology. Blockchain functions automatically, performing the obligations of a contract without the need for further human intervention. In theory, this model works well for most contracts. Vending machines are often used as an example to help understand smart contracts. In the example of the vending machine, the contract is to dispense a drink or a snack in exchange for a payment, and (other than making the payment) no human intervention is necessary. The contract is a simple one, with few opportunities for non-performance. As a contract becomes more complex, however, there are more opportunities for a malfunction that would impact performance.
For example, if the vending machine dispenses coffee, it typically will offer the buyer several options: sugar, creamer, sugar and creamer, or neither. Does the buyer want their beverage extra strong? Multiple options increase the complexity of the transaction, and with that complexity come more chances for the transaction to fail. AI can identify the likely causes of a malfunction, and the likely results of that malfunction, and direct the system to deal with it.
2.4.3 Reliability of outcomes
Blockchain is a deterministic system, meaning that the result of a blockchain transaction depends entirely on how the system is programmed to act. To guarantee that the blockchain transaction will always function properly, all the potential variables that could affect performance must be considered, and a plan for dealing with them developed. While it is impossible to address or even be aware of every possible variable that could arise, AI can make informed predictions about the likelihood of certain outcomes arising and formulate an appropriate reaction that would mimic what a person would (in theory) do in the same situation.
Section 3 – Commercial uses in practice
Smart contracts are commonly used across routine commercial transactions. They are best suited to transactions which result in payment of funds upon certain triggering events, or for the imposition of financial penalties if certain objective conditions are not satisfied; in either case, human intervention is not required. Some ‘real-life’ examples include the following:
3.1 Purchase and sale of goods
Contracts for the purchase and sale of goods lend themselves to automation as they follow a series of events that can be easily programmed where the buyer orders the goods, the seller ships them if they are in stock, and payment is made. This is the simplest form of contract. The parties may agree to add additional terms and conditions to meet varying contingencies, but the essential steps of the contract are the same even though it is a smart contract. There is no need to have a human process the order, or to go through the accounting or bookkeeping steps necessary to make a payment. The contract is performed, in essence, by the contract itself.
AI is useful in conjunction with the smart contract for the sale of goods. At the preliminary stages, AI can help make better judgments and predictions, making use of data such as buyer preference, purchasing cycles, and the percentage of shipped goods that are rejected or returned. This will make the process of selling and dealing with potential customers more efficient and allows the smart contract to be tailored more effectively to customer demand.
AI is also effective as a way of monitoring contract performance without the need for human intervention. When difficulties arise, an AI system can identify and address the difficulty by examining datasets or uncovering trends or risks. This monitoring capability will allow the use of smart contracts in a wider array of transactions, ultimately enabling more sophisticated analysis.
3.2 Real estate
AI is a powerful tool that is well suited for contract drafting and review. This is especially true in real estate contracts. Real estate contracts are complex documents made up of multiple, intricate, and interlocking terms and contingencies. AI allows rapid review, analysis, and explanation of these terms and contingencies, saving all parties to the contract time and expense.
Once the terms of a transaction are agreed, a smart contract can be used to eliminate or streamline real estate transactions. For example, a mortgage applicant could have their mortgage approved immediately upon the necessary documentation being submitted. Any credit checks or title searches may be done immediately, without needing to wait for human-initiated record searches. When that mortgage is approved, payment would be made to the seller, the necessary documents would be prepared, and the title transferred. Instead of a closing that takes place weeks after an offer is made, a smart transaction may be completed in a matter of hours. Digitizing real estate transactions removes the need for intermediaries to approve transactions, enabling buyers and sellers to transact directly in much less time and with fewer associated costs.
3.3 Financial transactions
Financial transactions, including fund transfers, are also well-suited to smart contracting. Once given conditions are met, the transaction is completed and recorded (eg, use Bitcoin to pay funds to Y as soon as the signal is received that all conditions of the transaction have been satisfied). By eliminating the need for processing costs and third-party intermediary intervention, this can reduce the costs and fees associated with transfers of funds. The blockchain technology used enhances efficiency and minimizes or eliminates security concerns.
A smart contract may not be able to detect indicia that give rise to a suspicion of fraudulent or criminal activity, because it is designed to react in a particular way to particular data. AI can be used to detect and analyze indicia of possible wrongdoing, such as data anomalies or suspicious relationships that in themselves are not illicit, but that merit further investigation.
3.4 Insurance claims processing
Insurance claims are increasingly being processed as smart contracts. When a predefined event occurs, such as a flight cancellation or natural disaster, the smart contract will automatically trigger payments without the need for manual claims processing. AI can enhance the claims process by analyzing historical data and patterns for more accurate predictions of claims trends, and can thus assess risks more accurately. The time for claims processing and settlement is reduced, the possibility for human error is minimized, and, perhaps most importantly, customer satisfaction is improved by quick and fair payouts.
3.5 Supply chain management
Supply chain operations are optimized by using smart contracts to automate processes such as inventory management, tracking shipments, and verifying the authenticity of goods. AI can enhance the operawtion and development of contracts by predicting demand, identifying potential disruptions, and optimizing logistics routes. This integration ensures that all parties have real-time access to accurate data and reduces delays, improves transparency, and enhances efficiency. The entire supply chain process is streamlined.
3.6 Energy and utilities
The energy sector has been transformed significantly in recent years. Energy providers are placing an increased emphasis on efficiency, reliability, and renewable sources. Energy sources such as solar power or biolmass have moved to being essential components for providing energy. Blockchain smart contracts have become a key technology that helps to address the limitations of the traditional energy grid and facilitate the growth of a more diverse, decentralized, and sustainable system. Self-executing programs, stored on a blockchain, automate transactions and processes, and are a fundamental shift in how energy is produced, distributed, and consumed.
Smart contracts have shown great promise in enabling transactive energy. Transactive energy refers to a system where energy can be traded directly between parties without the need for a traditional utility distribution system. This is particularly relevant for renewable energy sources like solar and wind, where generation can be intermittent and localized. Smart contracts are able to certify the origin of the energy,and can process transactions at high speed. There will be a permanent record of every exchange in an unalterable ledger. This not only enhances the security and transparency of transactions, but it also empowers individual consumersand encourages them to become active participants in the energy market through peer-to-peer (P2P) energy trading.
There are benefits to integrating smart contracts into the energy sector:
storing contract data on a blockchain provides a new level of security. The blockchain is resistant to editing and manipulation;
the automation inherent in smart contracts increases speed and efficiency, which in turn allows transactions to be processed as close to instantly as is possible, without the need for manual intervention;
the decentralized nature of these systems improves reliability by eliminating the single points of failure that are common weaknesses and vulnerabilities in traditional grids; and
removing the need for a central intermediary decreases the costs for all parties involved.
While smart contracts have immense potential for use in the energy sector, there are significant challenges to overcome. The code used to write these contracts is complex, and specialized developers are required. Any glitches or errors in the code can be difficult to locate, and even more difficult to fix, once the system is deployed. Additionally, as a relatively new technology, there are still unknown security vulnerabilities that will arise and that must be addressed to protect against cyberattacks and energy theft.
While some states have begun to pass legislation that grants smart contracts legal enforceability, the uncertainty and the lack of standardized regulations and legal frameworks also poses a hurdle. Despite these obstacles, the combination of blockchain, smart contracts, and a decentralized grid offers a path toward a more resilient, efficient, and sustainable energy future.
Section 4 – Legal status and liability
The legal status of smart contracts and contracts generated by AI remains unclear. Legal opinion on the status of smart contracts varies, with some commentators arguing that smart contracts are not really contracts at all but merely represent computer programs that provide a convenient mechanism for performing an existing agreement. Some jurisdictions may recognize smart contracts as legally binding, while others may not. At the other end of the spectrum, others have argued that it is possible for AI to generate an offer for a contract, and that AI could likewise indicate assent to that offer (eg, by shipping goods that were ordered by the offer). Such a contract could, according to this analysis, be valid in some circumstances, such as creating a contract for the sale of goods under the UN Convention on the Contracts for the International Sale of Goods (CISG).
Any discussion of smart contracts will show how easily difficult legal issues can arise. For example, since the agreement is memorialized in code, how can the parties be assured that the technical code captures their intentions, or indeed is accurate? In theory, the code would be written to be the record of the parties’ intentions regarding the agreement in much the same way that a document written by a human is a record of an agreement.
The written agreement, however, is in a language understood by the human parties involved. Code is a language largely decipherable only by computers. This has led some to conclude that smart contracts are just as susceptible to ambiguities as contracts written in a human language, because ‘technical facts depend on socially determined ones’ with the added pitfall that these ambiguities are hidden. See All Smart contracts are ambiguous, Journal of Law and Innovation, October 2019, James Grimmelmann.
Given that smart contracts are written by programmers rather than lawyers, the interpretation of how the contract should look may be based on the programmer’s understanding of how the contract should operate and how that should be written in code. Errors or bugs in the code could lead to (potentially costly) disputes or risk security breaches. The parties and their lawyers can check whether the code correctly represents their intentions; however, there is a risk of miscommunication in terms of what was actually agreed to, and potential errors may not become evident until the execution phase of the contract. For example, in some contracts, the terms used are subjective (eg, ‘acting in good faith’), but this is often a question of degree, which is difficult to translate into a code, or may be interpreted in a certain way by a programmer.
4.1 Valid contract formation
The basic rule of contract law is that a valid contract is created when an offer is made, that offer is accepted, and some form of valuable consideration is exchanged between the parties. There is a fourth element whose absence will make any purported agreement a nullity, even if offer, acceptance, and consideration are all in place. That fourth element is capacity. If any of these components are lacking, there will not be a valid contract, and the concepts are harder to apply where no human is involved.
The key question is: does AI have the legal capacity to enter a contract? For example, do computers have the legal capacity to make a contract? ‘Capacity’ has traditionally been defined in terms of a natural person: as a general rule, an adult natural person has full legal capacity to make a binding contract. Non-natural ‘persons’ (eg, legal entities, such as corporations) also have the capacity to make binding contracts, but that capacity is necessarily exercised through human actors.
Existing definitions of capacity do not include computer programs, and queries arise around the ability of AI to form an intention to create legal relations. Another consideration is whether the established legal defenses to a contract (eg, fraud, unconscionability, and mutual mistake) will ever be applicable. While there are ongoing efforts to impose criminal liability on the authors of AI algorithms that perpetrate fraud, the issue of defending a civil lawsuit for breach of a smart contract by using traditional contract defenses does not appear to have arisen.
4.2 Risks and liabilities of smart contracts
Smart contracts also pose additional risks alongside the issues of valid contract formation and performance. For example, contracts typically are drafted in final form after the parties have reached an agreement, or an understanding of how the agreement will look. In the case of a smart contract, the negotiations are completed, and the offer and acceptance would already be in place when the smart contract algorithm code is developed.
The parties should also consider their obligations under the contract, including the following:
- Who is liable if the code is inaccurate?
- Are any limitations to liability permitted – have these been discussed between the parties, and have exclusions or monetary caps on liability been negotiated?
- How will flaws or errors be detected?
- If a smart contract and an AI malfunction crosses jurisdictional borders, the laws of which jurisdiction will apply?
- What mechanisms are in place for dispute resolution?
- How will updates or modifications to the contract be handled?
Inaccuracies or omissions in the code may not become apparent until the contract has been in operation, and determining legal responsibility can be complex. Continual ‘human monitoring’ and oversight of contract performance and AI models may be necessary. To carry out such oversight, lawyers will need to consider the terms agreed regarding the proposed transactions.
4.3 AI and contract management
The most immediate application of AI to contracts is in the management of contracts. For example, a company that deals with a high volume of similar contracts will find AI useful for keeping the different, but still very similar, contracts separate from one another - eg, by identifying specific contract types, extracting key data and highlighting terms, rights, and obligations. AI is also being used for contract review and other matters of due diligence given its ability to analyze large volumes of data faster than manual checks.
Section 5 – Risks of enforcement action
Much of the legal activity and discussion in the United States regarding AI has, to date, been focused on issues relating to intellectual property and tort liability. Discussion of contractual liability focuses on liability for breach of warranty by the developer of an AI system, and contract litigation involving AI largely involves disputes between developers and customers.
States are only just beginning to catch up with the technology. Some have enacted legislation that addresses smart contracts specifically. For example, in Illinois, the Blockchain Technology Act (205 ILCS 730/) provides that a smart contract ‘may not be denied legal effect or enforceability solely because a blockchain was used to create, store, or verify the smart contract.’ Evidence of a smart contract may not be excluded solely because a blockchain was used to create, store, or verify the contract.
An Arizona law on signatures and records secured through blockchain technology, smart contracts, and ownerships of information (Az Rev Stat 44-7061) makes explicit the validity of smart contracts. The law provides that ‘[s]mart contracts may exist in commerce. A contract relating to a transaction may not be denied legal effect, validity or enforceability solely because that contract contains a smart contract term.’ The statute defines a ‘smart contract’ as ‘an event-driven program, with state, that runs on a distributed, decentralized, shared and replicated ledger and that can take custody over and instruct transfer of assets on that ledger.’ The term ‘with state’ is not defined.
Additional resources
Related Lexology Pro content
How-to guide:
Understanding the risk of negligence claims when using AI
Panoramic:
Reliance on information posted:
While we use reasonable endeavours to provide up to date and relevant materials, the materials posted on our site are not intended to amount to advice on which reliance should be placed. They may not reflect recent changes in the law and are not intended to constitute a definitive or complete statement of the law. You may use them to stay up to date with legal developments but you should not use them for transactions or legal advice and you should carry out your own research. We therefore disclaim all liability and responsibility arising from any reliance placed on such materials by any visitor to our site, or by anyone who may be informed of any of its contents.