Our research looks at the role lobbying plays in influencing artificial intelligence (AI) policy in Canada, especially during the agenda-setting and problem definition stages. Our research uses Critical Discourse Analysis (CDA) and policy agenda-setting theory to analyze federal parliamentary committee hearings and to uncover the underlying power dynamics of these stakeholder engagement venues. This research highlights a significant lack of representation of marginalized voices in this public forum, creating additional social exclusion for these groups in AI policymaking. As a result, policy recommendations stemming from these meetings did not properly account for AI-related risks felt by marginalized communities. Our findings show that lobbying groups use specific discursive strategies to further their self-interest, and that negativity bias strongly influences policymakers, prioritizing AI-related risks over benefits. Our findings contribute to the literature on social inclusion, lobbying, and AI governance. Thus, we emphasize the need for more equitable representation in AI policy discussions.