Unions Push AI Guardrails as 70% of US Agencies Adopt Generative Tools

Nearly 70 percent of U.S. government agencies now use generative AI, according to an Ernst & Young survey, and public‑sector unions are beginning to lock in safeguards against surveillance and automation‑driven layoffs. The Decoder reports that recent wins give small‑business leaders a preview of the labor‑policy pressure building around workplace AI.

What happened

The Decoder notes that an Ernst & Young survey finds “nearly 70 percent of US government agencies already use generative AI.” In response, unions across the country have secured concrete protections:

• After an October strike vote, IFPTE Local 21 in Oakland won a commitment to provide retraining programs instead of AI‑related layoffs.
• In Pennsylvania, Governor Josh Shapiro guaranteed SEIU Local 668 “human oversight” during a pilot of OpenAI’s ChatGPT Enterprise.
• The Transport Workers Union in Ohio secured “veto power over autonomous vehicles” for its members.
• California Governor Gavin Newsom banned “AI surveillance software, often called ‘bossware,’ along with automated firings” in September.
• New York has protected state employees from AI displacement by law since late 2025.

Labor researcher Kung Feng of UC Berkeley is quoted in the piece as saying employers are pushing automation to cut costs. Union leaders—including AFL‑CIO President Liz Shuler, Lorena Gonzalez, and TWU’s John Samuelsen—are backing AI safety systems that “help workers but insist on mandatory human control over every AI application in the workplace,” a claim the Decoder attributes to the Guardian.

Why it matters for small‑business owners

While the wins are in the public sector, the patterns are directly relevant to small and mid‑size firms that are also adopting generative AI to boost productivity. First, the “nearly 70 percent” adoption rate signals that AI tools are no longer niche; they are becoming baseline infrastructure, which means the same surveillance and automation risks that motivated union action are likely to appear in private workplaces. Second, the specific guardrails—retraining guarantees, mandated human oversight, veto rights over autonomous systems, and outright bans on bossware—show the kinds of policies that can pre‑empt labor disputes. Small firms that ignore these emerging expectations may face employee pushback, turnover, or even future regulation that mirrors the state‑level bans already in California and New York. Finally, the union‑backed emphasis on “mandatory human control” provides a practical framework: if a tool can make a hiring, firing, or scheduling decision without a human check, it becomes a liability. By adopting similar safeguards now, owners can avoid retrofitting policies under pressure later.

What to do this week

1. Conduct a quick audit of any AI tools your team uses to identify whether they collect employee performance data or make autonomous decisions; flag any “bossware”‑type features.
2. Draft a one‑page “human‑in‑the‑loop” policy that requires a manager to review and approve any AI‑generated recommendation that affects pay, scheduling, or termination.
3. Schedule a 30‑minute training session with staff to explain the new policy and gather feedback on any AI‑related concerns they have.

Sources

• Nearly 70 percent of US agencies use generative AI, and unions are finally catching up with guardrails — The Decoder