Alberta MLA’s AI claims misrepresent how Canada’s charities actually work

Canada’s non-profit sector is used to scrutiny. So, so much scrutiny. 

Charities file annual returns, undergo audits, report to funders and arguably operate under the most explicit, stringent transparency requirements in the country. 

Which is why Alberta MLA Nate Glubish’s recent post framing Canada’s charitable sector as wasteful is so disheartening and misleading.

Glubish’s post (which reads like it was written by AI, but that’s another column) claims that AI has identified “round-trip money flows,” “zombie organizations,” and “ghost capacity” across the charitable landscape – all terms that sound alarming yet reflect that Glubish appears to have little idea how charitable funding works or how non-profits operate.

He offers zero evidence that any of these so-called wasteful organizations exist, only that they might.

But let’s dig a little deeper into these terms.

Round-trip money flows: Federated charities regularly transfer funds between national and local chapters.

Zombie organizations: Project-based non-profits routinely dissolve when their mandate is complete, which is a sign of good governance, not waste or misconduct.

Ghost capacity: Volunteer-run organizations have no employees by design. 

None of these things are new, hidden or suspicious. 

The suggestion that these normal patterns indicate systemic waste risks eroding public trust in organizations that deliver critical, essential services. These are the same organizations that governments rely on to fill gaps in social infrastructure, such as food programs, housing supports, crisis response and community health. 

Undermining them ultimately harms the communities Glubish was elected to serve.

AI can surface correlations in large datasets – and hey, the massive but incomplete dataset provided by law to the government by charities and non-profits is an easy one to target – but AI can’t determine whether those correlations are meaningful, legal or harmful. 

It can’t distinguish between a legitimate transfer within a national charity and an improper one. It can’t tell the difference between a volunteer-run organization and a fraudulent one. It can’t validate its own findings.

Human expertise, including sector expertise, is essential. Otherwise, we’re handing over accountability to the machine – do the accused get to plead their case in such a system? Is that true accountability?

Generating numerous false flags for things that are normal is a massive waste of time, money and will ultimately lead to witch hunts. 

There are already numerous oversight mechanisms, such as the CRA Charities Directorate, mandatory T3010 filings, audited financial statements, provincial registries and extensive funder reporting. 

Glubish’s claim that analyzing charity networks involves “3.2 septillion combinations” is mathematically true, if you assume that every charity gives money to every other charity in a five-step chain. 

We know real charity networks are sparse, not fully connected. This kind of numerical inflation is designed to make the problem seem unsolvable without AI, which, again, is misleading.

And a small but infuriating contradiction: Glubish claims that the system would not use personal data, while also describing entity-matching across systems, which is a process that typically requires personal data. This does not give me confidence that the team understands privacy, governance, and data handling for vulnerable people. 

The issue here isn’t a lack of oversight in charitable systems. It’s that the federal and provincial governments have not modernized the systems that support it. 

It’s left this massive, public dataset incomplete, which will also result in more false flags.

Better data infrastructure would help everyone – regulators, funders, and charities alike – but only if it’s built with an understanding of how the sector actually operates. 

We’ve covered this issue extensively. We’ve also covered numerous charitable and non-profit organizations on the cutting edge of technology adoption. 

This is where AI should be. Not trying to look backwards to ferret out waste with a faulty premise.

No one is opposed to innovation here. Non-profits literally run on zero margins and are always looking for ways to save money and prevent waste. Many have gladly adopted the latest AI tools where it makes sense to do so. 

There is no evidence that fraud and waste are occurring at a massive scale in the charitable sector. When fraud does happen, it’s usually caught by the systems already in place.

Fix the data infrastructure to deliver real, complete data, making it easier for charities and non-profits to report to the government, and you’ll help prevent future fraud.

Canada’s charities are not a problem to be solved by technology. They are partners in delivering public good. Any effort to strengthen oversight should start from that premise.