The Ontario government is accelerating its push into artificial intelligence, with new data showing the Ontario Public Service (OPS) now has the highest usage of Microsoft Copilot chat tools in the country. Internal presentations obtained through freedom-of-information laws reveal that more than 15,000 provincial employees are using Copilot every week, generating over 120,000 views on the internal “Copilot Chat InsideOPS” portal — far more than any other jurisdiction in Canada.
With strict rules banning public servants from using AI chatbots like Google Gemini or ChatGPT due to privacy risks, Microsoft Copilot has become the only approved generative AI tool widely available to government staff. Ontario’s Minister of Public and Business Service Delivery and Procurement, Stephen Crawford, says early results show significant productivity gains.
“From what I understand, the average public servant is saving almost three hours per week,” Crawford told Global News. He added that hundreds of AI “leaders and pioneers” across government have been selected to test more advanced AI tools as part of the rollout’s pilot phase.
Government documents show Ontario is currently in the first stage of a three-part AI strategy. Phase One focuses on building the foundation and identifying practical use cases; Phase Two envisions developing specialized AI tools that could boost productivity by 20 per cent; and Phase Three — labelled “Industrialize AI” — imagines the technology being deployed across the entire public sector at scale.
Crawford said the broader shift toward AI won’t eliminate public service jobs but will instead reshape roles and free staff to focus on more valuable work. “There will be job transformation,” he said. “I don’t see job loss — I see job enhancements.”
So far, the examples are modest but illustrative: AI is being used for media monitoring, drafting news releases, and summarizing large documents. But industry experts say the potential is far greater. Cam Vidler, co-founder of AI firm Authentica and former chief of staff to the finance minister, says emerging “agentic AI” systems — AI agents programmed with precise rules — could handle complex administrative tasks that human staff currently process manually.
He points to situations like the province’s $200 rebate program ahead of last year’s election, which triggered thousands of exceptions due to outdated addresses, returned mail, or payments sent to people who had died. Those cases required human intervention because legacy IT systems couldn’t handle the complexity. Vidler says agentic AI could eventually manage such exceptions automatically and reduce the need for large call centres.
Crawford acknowledged Ontario is still years away from deploying AI at that level, particularly since Copilot remains the only approved tool for most employees. But he expects progress to accelerate. “It’s hard to say exactly when we scale up,” he said. “But if I had to speculate, I would say within a few years.”
Vidler cautioned that adopting AI across government should be gradual and strategic. Rather than launching broad AI policies, he suggests starting with targeted areas where automation can reliably reduce errors and free up staff time.
“You don’t have to decide to do AI on everything,” he said. “Pick a couple of low-hanging fruit examples — things you know are prone to errors, and that you don’t want your people spending time on.”
Ontario’s AI expansion marks one of the most significant technological shifts in the province’s public service in decades — and the government appears prepared to push much further as the tools evolve.