Survey Finds AI Safety Documents Lack Clear Accountability

A TRADESAFE survey of 514 U.S. workers found most lack a formal policy defining who answers for AI-assisted safety content errors.

Updated on Nov. 5, 2026 in Artificial Intelligence

Isometric editorial illustration of a lone steel pallet-rack upright and a hanging yellow safety chain in an empty concrete warehouse bay. — A TRADESAFE survey of 514 U.S. workers found only 18% say their workplace has a formal policy defining who answers for AI-assisted safety content errors. AI Illustration. Upload story photo >

Live Poll

Would you trust a workplace safety procedure more if a qualified person reviewed it?

In a September 2026 survey, TRADESAFE polled 514 U.S. workers in manufacturing, construction, warehousing, logistics, and the skilled trades about AI use in safety documents. Only 18% said their workplace has a formal policy defining accountability for AI-assisted safety content errors.

Why it matters

Half of respondents said their organization uses generative AI to write safety documents, yet only 49% said outputs are independently reviewed before use. Herbert Post said the central problem is the absence of defined ownership for errors in AI-drafted safety materials.

Of 399 respondents with visibility into safety document creation, 51% said their organization uses generative AI to write them, while 36% said they had personally seen an AI-generated safety document with an outdated requirement, vague instruction, or missing hazard step. Leaders reported far more visibility into document production (76%) than frontline workers (31%), and the survey does not report how many of the flawed documents led to actual harm.

The players

TRADESAFE

TRADESAFE is the organization that conducted the September 2026 survey of workers on AI use in safety documentation.

Herbert Post

Herbert Post is the spokesperson quoted in the survey coverage arguing that defined ownership of AI-drafted safety document errors is missing.

The details

Respondents were shown a safety procedure presented as written by a person, drafted by AI and then reviewed, or drafted by AI with no review. In that randomized scenario, 56% blamed company leadership or the safety department for a serious injury caused by an unreviewed AI draft, while 9% blamed the employee.

September 2026: TRADESAFE conducts the survey of 514 U.S. workers.

Big Picture

Written safety procedures in U.S. workplaces are already subject to employer responsibility rules, and this survey shows generative AI drafting is entering that same documentation without clear accountability. Only 18% of workers reported a formal policy defining who answers for AI-assisted errors, suggesting the oversight gap predates any specific rule for AI.

Workers in manufacturing, construction, and the trades may be following safety procedures that were drafted by AI without a clear reviewer. Asking who approved a given safety document and whether it was independently checked is a practical step for employees who rely on those instructions.

The takeaway

Safety documents are only as reliable as the review behind them, and the survey suggests that review is missing for most AI-drafted material. Workers can ask who owns a procedure's accuracy before relying on it.