AI And Automation At Construction Sites: Powerful Tools Or New Source Of Liability

What to Know

• AI boosts construction safety and efficiency but raises new liability risks.
• Construction firms need clear AI governance and risk controls.
• AI-driven jobsite gains must be balanced against legal exposure.

Firms that deploy AI and automated equipment can potentially gain advantages in safety, speed and cost, but the legal infrastructure to support it—training, more specific contracts, insurance and processes for assessing losses and the preservation of evidence—must be built deliberately and not retrofitted after a loss.

Emerging technologies are nothing new. From electricity to computers to the internet, each major wave of innovation has arrived with uncertainty, quirks and growing pains. Yet history shows that these technologies ultimately become embedded in society, fundamentally changing how we live and work.

Time and again, those who resist or ignore such advancements, or blindly use them without the proper education or training, risk being left behind. Those who embrace advancements and learn to use them effectively and manage their risks are best positioned to lead and succeed.

The same holds true with AI and automation in the construction industry; owners, contractors and design professionals who effectively use AI tools will outpace those who do not, and those who use them carelessly will expose themselves to unfortunate liabilities.

AI and Automated Tools on Project Sites

AI and automation at construction projects have moved from pilot programs to common use, especially for progress tracking, scheduling, drone monitoring and inspections. Some projects use semi-autonomous equipment, like excavators, bulldozers for grading, trucks and loaders for material transport and building equipment such as bricklayers and rebar tying machines, among others.

These AI tools are being adopted and used because they promise efficiency, better hazard detection, faster data processing, improved coordination and lower costs across complex projects. Indeed, a 2022 a study, found autonomous construction robotics can reduce repetitive labor by up to 90% and cut worker exposure to hazardous conditions by 70% (Brosque and Fischer, Construction Robotics, June 2022).

At the same time, there are still no OSHA standards for the robotics industry, which means disputes regarding automated equipment and AI-enhanced processes will still be determined by good old fashioned workplace safety protocols, contractual provisions, common law negligence standards and products liability laws.

Unfortunately, most statutes, codes, construction contracts, insurance placements and other jobsite protocols were created exclusively for human decision makers, not for AI-assisted processes or robotic systems that can malfunction, misclassify hazards or produce hidden errors.

Liability Issues

The types of liabilities that may be encountered on construction sites due to AI or automated equipment can be placed in three categories: (1) personal injury or property damage, (2) construction defects and (3) delay damages.

The problem with these claims in the context of AI-assisted technology and automation is that the causal chain is rarely clean, and existing contract forms do not specifically address AI or autonomous systems. The American Institute of Architects (AIA) forms do not have any provisions addressing liability for AI or autonomous equipment. This is problematic because the chain of liability can be complex and involve multiple parties, including the owner, prime contractor, subcontractors, equipment manufacturers, technology vendors and software companies.

Compounding the problem is the fact that equipment manufacturers, technology vendors and software companies are very adept at including limitations of liability and disclaimers in their purchase and supply contracts, and historically, these contracts are often overlooked or not thoroughly negotiated by contractors or others at the jobsite.

This hypothetical illustrates the liability concerns: An autonomous hauler misinterprets a worker’s path due to a sensor glitch and fails to stop, striking and injuring the worker in a mixed traffic zone. The contractor gets sued for personal injury claims and OSHA citations are issued, and the owner also gets sued.

The vehicle manufacturer argues that the contractor did not properly segregate pedestrian and vehicle traffic or provide an updated site plan to the AI system when confronted with indemnity claims from the contractor and asserts the disclaimers and liability limitations in the contracts at issue, leaving the contractor and owner without recourse.

Further, the disruptions created by the accident, including the repair and/or replacement of the hauler and subsequent grounding and investigation of other autonomous equipment at the site significantly delays the project, leading to disputed change orders and indemnity claims between the owner and the contractor.

The subsequent litigation drags on for years, with the insurers denying coverage and defense for the contractor and owner, leading to significant legal fees and expenses.

As demonstrated in the hypothetical, when AI‑driven or automated equipment fails on a construction jobsite, owners, general contractors, specialty contractors, equipment manufacturers and software providers may all face liability for personal injuries, construction defects and project delays.

Owners and contractors can be exposed to negligence claims for inadequate supervision, training or maintenance of AI systems, while manufacturers and AI‑software vendors may face products liability or negligence claims if design flaws, faulty algorithms or inadequate warnings contribute to harm. Specialty contractors and field crews may also be liable where they misuse, override or improperly operate automated systems that cause injury or property damage.

Safeguard Against AI and Automation Risk

While AI technology and automated machinery used at construction sites may be new and complicated, the strategy to head off risks associated with the technology is not new or complicated. The best strategy is to ensure all these scenarios are addressed in the relevant contracts.

On many projects, the most important claims may be contractual rather than tort-based. If an AI-related failure causes rework, delay or downstream trade damage, parties will first look to the prime contract, subcontracts, consultant agreements and vendor terms to determine who accepted the risk and who promised to indemnify whom. As a result, common contract clauses on means and methods, delegated design, site safety, and correction of work that are not specific to AI may end up carrying a much heavier burden than they were designed to bear.

As a result, practitioners must carefully address AI and automation risk in contracts, insurance and safety programs rather than assuming technology alone reduces liability exposure. The most practical response is to revise project and vendor documents before the loss occurs.

Contracts should require the disclosure of when AI and automated systems will be used and define AI systems and automated equipment broadly enough to capture software, robotics, drones, machine-learning tools and third-party platforms embedded in a vendor’s offering.

The contractual provisions at issue should also specify who selected the technology, who is authorized to operate it, what training is required, and whether human review is mandatory before an AI output drives a safety, design, or payment decision. These provisions can be inserted into agreements in a specific section addressing AI technology and automated systems or may be provided for in an appendix or exhibit that is referenced in the standard contractual representations and warranties, liability disclaimers and/or indemnity provisions.

It is also important for owners to address AI related issues and the use of automated equipment in bid packages and draft contracts included within those bid packages so that contractors know in advance what they will be required to disclose, what to negotiate in their vendor contracts, when they can use AI and automated equipment and, if so, what precautions and procedures they must undertake in advance.

Several provisions in these agreements and general conditions deserve special review and modification, whether they be in the construction contracts between the owner and prime contractor, the architect’s contract, specialty contractor agreements or, more importantly, manufacturer and vendor supply and purchase contracts as follows.

1. Add definitions of “AI technology” and/or “automated equipment” to clarify what the new contractual provisions cover.
2. Include disclosure clauses requiring parties to identify AI tools and autonomous systems used on the project, and whether such technology and equipment fall into the contractor’s “means and methods.”
3. Include human-in-the-loop clauses stating that AI output only supplements professional human or supervisory judgment and does not replace it.
4. Add modification of safety clauses to require written procedures for operating, monitoring and overriding AI controlled equipment, drones or machines.
5. Specifically address which parties are responsible for certain failures of AI technology and equipment, i.e., software failures, data input errors, omissions, misuse of the technology or equipment at the project site or design flaws or defects caused by the architect’s use of AI technology.
6. Require data-retention and cooperation clauses for the preservation of logs, alerts, updates and incident records, potentially with procedures identified to access and preserve such information and data.
7. Include carve-outs for damage limitations or caps that may be applicable to AI technology or automated equipment that reflect project-level exposure rather than a cap limited to a small multiple of software subscription fees.
8. Tailor indemnity clauses to allocate responsibility for bodily injury, property damage, and rework arising from defective or misused technology. Also, the inclusion of indemnity provisions for intellectual property claims related to the misappropriation or infringement of any AI technology or automated equipment used at the project.

Without these provisions, a contractor or owner may discover that the AI vendor’s boilerplate language disclaims consequential damages, limiting liability to the fees paid under the software subscription, and disavowing any warranty that the system is fit for safety-critical construction.

AI-related construction disputes also add an evidentiary problem to the usual fight over fault because these systems involve complex interactions among humans, machines and changing site conditions, which makes incident reconstruction difficult. That complexity becomes worse when the most useful evidence is in vendor-controlled logs, sensor data, update histories or cloud-hosted data that may not be preserved automatically after an event.

For that reason, counsel should expect aggressive discovery fights over access to training assumptions, alert histories, software updates and override records. A party that does not preserve those materials after an incident may lose the best evidence of what the system saw, what it recommended, and whether a human accepted or rejected that recommendation.

Insurance: Have a Coverage Plan

Insurance can soften the impact of AI failures, but only if the coverage is applicable to the risk. Existing commercial general liability (CGL), workers’ compensation, builders’ risk, professional liability, cyber, and technology errors and omissions (E&O) policies may respond to parts of an AI-related loss, but none of these forms were drafted specifically for construction AI.

In situations where a bodily injury is caused by faulty software alerts or property damage results from erroneous automated measurements, coverage disputes are likely because the loss involves both physical and digital features.

Therefore, owners and contractors should discuss with brokers and coverage counsel whether: (1) software or cyber-related exclusions create gaps in coverage, (2) additional insured status locks in coverage from AI vendor policies, (3) professional-liability coverage extends to AI-assisted design or coordination work and (4) the policy conditions require immediate preservation of digital evidence after an incident.

Requiring technology vendors to carry technology E&O coverage in meaningful amounts is often as important as negotiating a strong indemnity clause, because an indemnity with no coverage behind it may have little practical value.

Practical Approaches

It is important for owners and contractors to take stock of all AI technology and automated equipment that will be used on their construction project before those projects commence. As mentioned earlier, there should be contractual obligations to identify and allocate risk for AI technology and automated equipment.

Additionally, a plan with action items and a numbered checklist is also a practical and appropriate way to deal with these issues:

Audit your technology before deployment—know what AI systems and equipment are on your site and who bears responsibility for each.

Update your contracts now—standard AIA forms or older in-house forms do not address AI liability; revisions and addenda as discussed above are necessary.

Demand vendor indemnification with teeth and expand liability caps to reasonable levels. Most vendor and manufacturer agreements are grossly insufficient for construction exposure.

Document human oversight—regulators will ask whether a human reviewed AI outputs before action was taken, and evidence of human oversight will be important in any litigation or arbitration.

Review your insurance program—coordinate with your broker to identify coverage gaps and new policy offerings that may be more specific to AI technology issues.

Train your people—liability exposure increases when workers misuse or over-rely on automated systems.

Owners and contractors should also have a process to address losses potentially caused by AI technology and/or automated equipment. A useful way to analyze an AI construction claim is to promptly and clearly document the loss, including any potential causes of the loss, and to have a process in place to investigate the loss, interview involved parties and preserve all evidence related to the loss.

The framework of any process and investigation should, at a minimum, focus on the following issues and questions:

Which party selected and configured the technology for the part of the project at issue?

Which party controlled the conditions under which the AI or automated equipment was used, including training, supervision, exclusion zones, and override authority?

What do the contracts say about means and methods, site safety, design responsibility, reliance, indemnity, and evidence preservation?

Did the failure stem from defective technology, bad input data, unreasonable human reliance, unsafe field conditions, or some combination of those causes?

Which insurance policy is most likely to provide coverage once the facts are characterized as personal injury, property damage, professional error or cyber-event.

This framework does not eliminate gray areas, but it does help to identify where liability for an incident may fall. In most disputes, liability will not be attached simply because AI was present; it will attach because one or more parties failed to supervise, document or minimize the risk in a way that matched the technology’s actual role on the project.

Conclusion and Takeaways

The most important lesson is that AI does not replace traditional construction-risk analysis. Instead, it intensifies the need for it. OSHA’s existing approach to robotics, the current state of construction technology and the structure of standard contracts may all lead to a legal analysis that is insufficient to identify liability. It is better to have more specific rules, processes and contracts to address these issues.

For lawyers advising owners, contractors, design professionals and trade partners, the immediate value lies in making the risk visible before the loss. That means auditing what AI is being used, revising the contracts that govern it to define it and allocate the risk, demanding better vendor contractual provisions to more fairly allocate such loss, implementing policies to preserve digital evidence, and making clear that human responsibility and accountability are still imperative.

Firms that deploy AI and automated equipment can potentially gain advantages in safety, speed and cost, but the legal infrastructure to support it—training, more specific contracts, insurance and processes for assessing losses and the preservation of evidence—must be built deliberately and not retrofitted after a loss.

Otherwise, the advantages of using AI technology and automation may be outweighed by increased potential costs and losses. 

This article originally appeared in Construction Dimensions magazine. Published with permission.