━ Overview
On a recent episode of EHS on Tap, host Jay Kumar sat down with R. Mukund, Founder & CEO of Benchmark Gensuite, to unpack the findings of Benchmark’s second annual EHS Benchmarking Report. The conversation moves past the headline number — 92% of EHS professionals now use generative AI — to the harder question underneath it: why injury frequency is still climbing even as that adoption curve flattens out at nearly universal.
This page expands on that conversation: what the underlying data shows about where AI adoption is actually landing inside EHS teams, why personal familiarity with the technology hasn’t yet translated into safer operations, and what it will take to close that gap.
━ Benchmark Gensuite Perspective
Two numbers from Benchmark Gensuite’s 2026 EHS Benchmarking Report should be read side by side. Generative AI use among EHS professionals has reached 92%, up sharply from the year before and now essentially universal. In the same survey, 45% of respondents reported an increase in workplace injury frequency over the prior 12 months, up from just 18% a year earlier, and 39% reported an increase in injury severity, up from 13%. AI adoption went up. So did the injury numbers. That’s the finding worth sitting with, because it isn’t the contradiction it first appears to be.
Almost all of that 92% adoption figure is concentrated in one layer of the job: drafting incident reports, summarizing safety data, writing procedures, cleaning up records. These are genuinely useful applications, and the time savings are real. But they are administrative tasks that happen after a decision has already been made, or in support of documenting one. None of them touch the moment where a risk is actually identified, assessed, or acted on before it becomes an incident.
That’s the layer where adoption is still early, and for good reason. Building AI into the actual business process, permit reviews, chemical approvals, hazard triage, is a different kind of undertaking than using it to help draft a document. It means trusting the technology with judgment calls that carry compliance and safety consequences if they go wrong, which is exactly why EHS leaders are moving more deliberately here than they did with personal, everyday use.
“Adoption is increasing on a personal level among EHS professionals — they’re seeing the value. What they’re now challenged by is figuring out how to implement it in the business process in an intelligent, value-added way that doesn’t increase risk.”
— R. Mukund, Founder & CEO, Benchmark Gensuite
What closes the gap isn’t more personal AI use — that curve is already close to flat. It’s AI that operates inside the workflow where the risk decision itself happens: analyzing incident text and photo data at volume to surface which hazards are clustering in which locations before they repeat, or working through a permit’s requirements the moment it lands rather than weeks later when someone finally has time. That’s a materially different kind of deployment than a drafting assistant, and it’s the one most organizations haven’t gotten to yet.
For EHS teams, the practical takeaway is to stop treating “AI adoption” as a single number. The 92% figure describes something that’s already largely solved. The real work, and the real opportunity to affect injury trends, sits in the much smaller percentage of organizations that have gotten AI embedded into the workflows where a risk decision actually gets made.
━ Platform context
Benchmark Gensuite delivered generative AI to its full user base in May 2025, ahead of most of the industry, which is part of why personal adoption inside its subscriber base tracks closely with the broader survey findings. But the platform’s more relevant work sits one layer deeper. Genny AI’s PSI Risk Advisor is built specifically to sift through large volumes of incident text and photo data to surface which significant risks and critical hazards are ticking up, and where, so that EHS teams aren’t relying on someone noticing a pattern manually across dozens of sites with very different risk profiles.
That same operational-layer approach extends to Genny AI’s Program Expert Agents. The Permit AI Agent walks a team through a compliance permit the moment it arrives, deconstructing requirements and building a compliance calendar, so a junior team member can get started immediately with an experienced reviewer validating rather than starting from a blank page. The Chemical Management AI Agent applies the same model to chemical approvals. In both cases, the goal isn’t to remove human review, it’s to move AI further upstream, into the workflow itself, rather than leaving it confined to drafting and summarization after the fact.
Benchmark Gensuite is also working with technology partners across computer vision and wearables so that these operational AI capabilities feed a single, unified data source, rather than creating the kind of disconnected data islands that make predictive insight harder to act on.
━ Key themes
━ What this means for EHS leaders
━ Next steps
The next 12 months of AI adoption in EHS should be treated as a workflow design project, not a continuation of the personal-productivity phase that’s already largely played out. Closing the gap between AI use and AI impact means being deliberate about where the technology sits relative to an actual risk decision, not simply how many people in the organization have tried it.
Start by mapping where AI already lives informally in your organization, mostly drafting and summarization, against where the highest-consequence risk decisions actually get made, permit reviews, chemical approvals, incident triage. Prioritize closing that specific gap before expanding AI use elsewhere. At the same time, work toward a unified data strategy across incident reports, wearables, and computer vision inputs, so that predictive insight isn’t undermined by disconnected data sitting in separate systems.
If you’re a current Benchmark Gensuite subscriber, the Program Expert Agents built for this exact gap, on permits and chemical management, are already available in your platform. If you’re evaluating EHS technology, this data is a useful benchmark for your own organization: if your AI use is still concentrated in drafting and summarization, you’re in the same position as most of the industry, and the opportunity to move ahead of it is open.
━ Featured executive
2026 EHS Benchmarking Report
Download the full findings behind this episode: The EHS Pressure Point, Benchmark Gensuite's second annual EHS Benchmarking Report.
Frequently Asked Questions
Why are workplace injury rates increasing in 2026 despite rising AI adoption?
What is the gap between personal AI use and organizational AI adoption in EHS?
Why is incident under-reporting getting worse among EHS teams?
How is total worker health becoming part of the EHS function?
Why are EHS teams taking on more sustainability and governance responsibilities?
What did Benchmark Gensuite's 2026 EHS benchmarking report find about workforce demographics and injury risk?
The report’s roughly 220 respondents, more than half at director level, pointed to workforce churn and generational turnover as contributing factors behind rising injury frequency and severity. As more experienced, long-tenured workers leave and are replaced by less experienced staff, jobs become more variable and less predictable, a combination the report ties directly to the increase in reported injuries.
SEE IT IN ACTION
See how enterprise EHS teams use our platform to manage risk, compliance, and safety outcomes at scale.
Note: Subscriber instances where you have an active registered user account are listed above.
If you need further support, please email us at
getHelp@benchmarkdigital.com
Please include your company name and registered email address so we can assist you.