Tax season 2026 brings new AI tools—and new AI scams. The CRA's Charlie chatbot got a $18M generative AI upgrade, but how do you know which AI tools to trust?

In 2024, Canadians lost over $638 million to fraud, with AI making scams harder to spot. Voice cloning now takes just 3 seconds of audio. Deepfakes fooled an Ontario man out of $12,000 using fake Trudeau videos. One Hong Kong company lost $25 million to a deepfake video conference.

This episode covers: CRA's AI chatbot accuracy and limitations, how voice cloning and deepfake scams work, who's most at risk (newcomers, seniors, everyone), protecting yourself from AI-powered phishing, what to share (and not share) with AI chatbots, why Bill C-27 died and what regulators are doing now, and practical steps to stay safe this tax season.

Scammers are ready. Make sure you are too.

Timecodes: 0:00 Introduction: Tax season meets AI 0:45 CRA's Charlie Chatbot: $18M upgrade explained 2:30 The AI scam landscape: $638M stolen in 2024 3:45 Voice cloning & deepfake threats 4:50 Who's most at risk 5:40 How to protect yourself 7:15 What regulators are (and aren't) doing 8:15 Practical steps for 2026 8:50 Final takeaway

Canada's AI ministry used AI to analyze 11,300+ public consultation responses—despite internal warnings it could "further undermine public trust." This episode unpacks what happened when the federal government turned to Cohere, OpenAI, and Anthropic to parse feedback on Canada's AI strategy, and why critics say the entire process was designed to prioritize industry over public concerns.

We cover the leaked November briefing note from Associate Deputy Minister Mark Schaan, the controversy around the 28-person industry-heavy task force, and why major issues like Indigenous consultation and environmental impact barely registered in the government's questions. From Kevin O'Leary's $70 billion Wonder Valley project on Treaty 8 land to the transparency questions around how AI actually analyzed your responses, this is the story of who really gets heard when Canada shapes AI policy.

The strategy drops any day now. By then, it becomes policy.

EPISODE TAGS/KEYWORDS

Canada AI strategy, Evan Solomon, government consultation, AI policy, Indigenous rights, digital sovereignty, OpenMedia, public trust, Ministry of AI, consultation process, Wonder Valley, Kevin O'Leary, Treaty 8, AI governance, transparency

EPISODE CATEGORY

News & Politics / Technology

TIMECODES

0:00 - Intro: AI analyzing AI feedback 0:45 - The breaking news from National Observer 2:15 - Internal government concerns revealed 3:45 - The bigger pattern: industry-captured consultation 5:30 - Wonder Valley: consultation without consultation 6:45 - The trust problem and transparency questions 7:30 - What to watch as strategy drops 8:15 - The takeaway: who really decides?

Canada is spending billions on AI infrastructure—data centres, compute capacity, sovereign clouds. But what if we're building the wrong thing?

In this episode, we examine a growing debate in Canadian AI policy. Daniel Wigdor argues that Canada's focus on data centres misses the bigger opportunity: building AI applications and companies. Mark Doble warns that billions spent on hyperscale infrastructure will soon face scrutiny. Meanwhile, globally, hyperscalers are spending $527 billion on AI infrastructure while generating only $25 billion in AI revenue—a gap that has investors nervous.

We break down the applied computing argument, Canada's IP drain problem (75% of AI patents go to tech giants), and whether Ottawa's infrastructure-first strategy will leave us as landlords in someone else's AI economy.

Plus: what to watch in 2026 as this debate intensifies.

TIMECODE TABLE OF CONTENTS:

0:00 - Introduction 0:45 - The Critique: Wigdor's Challenge to Infrastructure Spending 2:15 - The Applied Computing Argument 3:45 - The Counterargument: Government's Defense 4:30 - The Global Context: $527B Spending vs. $25B Revenue 5:45 - The Talent and IP Question 6:45 - What to Watch in 2026 7:30 - The Takeaway 8:15 - Outro