Budget 2026 announced Singapore's AI Mission for Advanced Manufacturing. The goal is straightforward: leverage artificial intelligence to build best-in-class manufacturing facilities, create first-in-world solutions, and drive broad-based productivity transformation across the sector, according to EDB. The A*STAR AI Centre of Excellence in Manufacturing has already supported close to 30 firms in developing and adopting AI-enabled solutions since its launch in 2024, and will expand its suite of common AI models to help more manufacturers.
Singapore already has the second-highest robot density globally at 730 robots per 10,000 workers, according to EDB published figures. The AI Mission is not starting from zero. It is building on an advanced manufacturing base that is already more automated than almost anywhere else in the world.
The question that has not yet been clearly answered for Singapore manufacturers adopting AI-driven production systems is this: when the AI makes a mistake, who is responsible, and does your insurance cover it?
How AI changes the liability picture in manufacturing
In a conventional manufacturing process, the source of a defect is traceable. A component fails its dimensional check because a machine was out of calibration. A batch fails its quality test because a raw material was substandard. A product does not perform to specification because a human operator followed the wrong procedure. In each case, the cause is identifiable and the liability path is clear.
When artificial intelligence is integrated into the manufacturing process, the liability picture becomes more complex in ways that most standard insurance policies were not designed to address.
Consider a manufacturer that has deployed an AI-driven visual inspection system to replace manual quality control. The system has been trained on thousands of images of conforming and non-conforming components. It classifies each unit as pass or fail in real time and releases conforming units to the next stage of production.
A batch of units passes the AI inspection and ships to the customer. At the customer's facility, the units fail their incoming inspection. The customer's production line is halted. The customer brings a claim against the Singapore manufacturer for the cost of the production stoppage and the delay to their own customer commitments.
The manufacturer's product liability policy covers bodily injury and property damage. No one was hurt. Nothing was physically broken. The units simply failed to meet the specification. The product liability policy is silent.
The question then becomes whether the AI system's misclassification is a product defect in the units themselves, a professional error in the design or implementation of the AI inspection system, or a service failure in the ongoing operation of the system. The answer determines which policy, if any, responds, and whether a gap exists.
The three liability questions that AI manufacturing raises
Who designed and implemented the AI system?
If the manufacturer developed its own AI inspection or process control system in-house, the liability for a system error sits with the manufacturer. If the AI system was supplied by a technology vendor, the liability may be shared between the manufacturer and the vendor. For the customer-facing claim, what matters is the manufacturer's own liability. The claim arrives at the manufacturer first.
Is the AI system's error a product defect or a service failure?
This question determines which part of the insurance programme responds. An AI system that was incorrectly trained, whose training data was not representative of the actual production environment, or that was not properly maintained as the production environment changed may represent a professional error. Standard product liability does not address any of these scenarios because the harm is financial, not physical. Manufacturers E&O insurance is designed for exactly this space: a product or service that failed to perform as specified or contracted, causing economic loss to a customer.
Does the AI system create an ongoing monitoring and updating obligation?
AI systems trained on historical data may degrade in performance as the production environment changes. New component geometries, new materials, new suppliers, or changes in the manufacturing process can all affect the accuracy of an AI inspection system trained on earlier data. A manufacturer that deploys an AI system and does not maintain it as the environment changes may be creating an ongoing professional liability exposure.
What the insurance programme needs to address
For a Singapore manufacturer operating AI-integrated production systems, the insurance programme needs to address three distinct exposures that standard policies do not automatically cover.
Manufacturers E&O for economic loss from AI system failures. When a product that passed an AI quality check fails at the customer's facility, and the claim is for financial loss rather than physical harm, Manufacturers E&O insurance is the policy that responds. It covers claims for economic loss arising from a product that failed to perform as specified or from a service associated with the product that was not delivered as contracted.
Technology E&O for AI system development and supply. For manufacturers that develop AI systems in-house and deploy them in customer-facing applications, or that supply AI-assisted inspection services to third parties, a technology errors and omissions endorsement addresses the professional services dimension of the AI system itself, separate from the products manufactured using it.
Cyber insurance for AI system integrity. AI manufacturing systems are connected. A cyber attack that corrupts an AI model's parameters, that poisons its training data, or that manipulates its real-time outputs creates a risk that is neither a conventional product defect nor a conventional cyber incident. Cyber insurance for a manufacturer using AI needs to address the integrity of the AI systems as part of the covered computer network, including operational technology systems in the production environment.
What to check before the next AI deployment
For a Singapore manufacturer that has deployed, or is planning to deploy, AI systems in its production or quality management processes, three questions are worth addressing specifically.
Does the current product liability policy cover economic loss claims from customers when AI-inspected products fail at the customer's facility? If not, does the programme include Manufacturers E&O cover that addresses this?
If the AI system was supplied by a technology vendor, what are the vendor's contractual obligations if the system fails? Does the manufacturer's own insurance respond to a customer claim regardless of the upstream vendor relationship?
Does the cyber policy cover the integrity of AI systems and machine learning models as part of the covered computer network, or does it address only conventional IT infrastructure?
You can read more about our approach to Manufacturers E&O insurance in our Insights post and about our product liability cover and professional indemnity cover on the products page.
If you are a Singapore manufacturer that has integrated AI into your production or quality management processes and would like to understand how your current insurance programme addresses the liability that arises when AI systems fail, we would be glad to work through it with you.
This article provides general information only. It is not insurance advice. Data on Singapore's advanced manufacturing sector sourced from EDB Singapore's published AI Mission for Advanced Manufacturing announcement and EDB published industry analysis. The claim scenarios described are illustrative and not based on actual cases. Policy availability, terms, conditions, and exclusions vary by insurer and product, and cover is subject to the full policy wording. Please contact TZY CO for advice on your specific situation.