Determining how to use has always been difficult; however, the increased use of AI is presenting an even more pressing challenge. 

A new report, Risk Recalibrated: The 2027 Executive Leadership Report on SIF, Technology, and the EHS Function, provides a snapshot of how organizations identify and manage serious-harm risk and the challenges they face.

EHS work is getting easier and harder at the same time

The data from Evotix and What Works Institute shows AI use has added speed, scale and new capabilities to information retrieval, analysis, training and reporting. Top use cases are:

• Chatbots and copilots for procedures or training rose to 70%, up from 44% last year.

• Predictive analytics for SIF risk increased to 49%, up from 24%.

While technology is making data collection and reporting easier, staffing constraints, retiring expertise, administrative demands and organizational complexity make other aspects of EHS work harder.

Priorities identified for the next 18–24 months rank SIF exposure tracking first and simplifying safety systems second. Human performance and learning rose to third, up from sixth in the previous survey.

"The conversation has shifted from whether to adopt AI to how to use it effectively in managing serious-harm risk," said Jonathan English, CEO of Evotix, in a statement. "Respondents are finding practical applications that make information easier to access and analyze, but turning those capabilities into prevention takes people who understand the work and can question what the information is telling them."

Human authority to challenge AI outputs varies

AI is beginning to inform safety decisions, but integration of expert human input hasn't caught up.

• Fifty-nine percent of respondents report that their organizations use AI outputs for safety decisions.

• Among that group, nearly all report human review before action (97%), but less than half (44%) say reviewers have explicit authority to disagree with the AI and document that disagreement.

• Concerns about AI's effects on safety decisions include fabricated or unsupported information that appears credible (58%) and AI replacing rather than augmenting practitioner expertise and judgment (34%).

"AI should help EHS leaders deepen their understanding of risk and augment their expertise, not outsource their judgment," said John Dony, CEO of What Works Institute, in a statement. "As AI use expands, organizations must preserve and prioritize their safety team's practical experience and critical thinking for informed decision-making."

SIF approaches are more established, but measurement gaps remain

With SIF exposure tracking and simpler safety systems topping future priorities, the findings highlight a continued focus on understanding where serious harm could occur and turning that information into preventive action. While SIF programs are more established, that progress hasn't solved the measurement challenge.

• Sixty-two percent of respondents describe their organization's SIF approach as established or fully integrated, compared with 37% last year.

• Less than one-third say safety metrics closely reflect the factors driving serious-harm risk (26%).

• Few say a majority of their safety KPIs measure effectiveness or control reliability (28%) rather than activity counts.

• Nearly all report safety to the board or executive team (97%), but more than half say reporting still focuses mainly on lagging performance (56%).