Ontario urgently needs 'guardrails' on public sector use of AI, say privacy, rights commissions

Province working on framework to 'ensure the transparent and accountable use of AI' across sector: ministry

cbc.ca
May 26, 2023

Ontario should put in place "effective guardrails" on the public sector's use of artificial intelligence technologies, the Information and Privacy Commissioner and the Ontario Human Rights Commission said Thursday in a joint statement.

The government must urgently develop rules so it can reap the benefits of AI technologies in an ethically responsible manner, the two bodies said in the statement.

"AI technologies have great potential to benefit society in terms of improved health, education, public safety, and social and economic prosperity," they wrote.

"However, they have also been shown to be unsafe when not effectively governed."

Setting rules for AI technologies
A spokesperson for the Minister of Public and Business Service Delivery said the government is developing a trustworthy AI framework "that will ensure the transparent and accountable use of AI" across the public sector.

That framework says the government commits to always being transparent, disclosing when, why and how AI algorithms are used, following rules to safely and securely apply those algorithms and protecting Ontarians' rights.

The human rights commission and privacy commissioner say they commend the foundation proposed in that 2021 framework, but it is "urgent for the government to establish a binding set of robust and granular rules for public sector use of AI technologies."

Privacy bill sets out rules on use of personal data, artificial intelligence
Before worrying about AI's threat to humankind, here's what else Canada can do
The two bodies say AI technologies have great potential to benefit society, but they can be unsafe when not properly governed and may unlawfully collect personal information.

Even with de-identified information, AI technologies can perpetuate biases and lead to negative effects on marginalized people or groups, they write in the statement.

As well, they say some AI systems can create flawed or inaccurate content, raising concerns about accountability.