The Ontario Securities Commission (OSC) recently released exploratory staff research, which found a substantial rise in the number of Canadian listed issuers referring to artificial intelligence (AI) in their management discussion and analysis (MD&A) filings.
The study delves into how often issuers refer to AI in their financial disclosures and whether they appear to treat AI positively or negatively, according to the OSC’s news release.
"This research underscores the evolving landscape of AI in corporate disclosures,” said Leslie Byberg, the OSC’s executive vice president for strategic regulation, in the news release.
The OSC shared that its staff who conducted this research described the 2024 filing year as a turning point in issuers’ overall outlook when mentioning AI, with a marked increase in negative sentiment. Before, most issuers tended to refer to AI positively.
The OSC staff also noted that issuers in the finance and information sectors generally have a more positive sentiment regarding AI, which diverges from the outlook of other industries that view AI more cautiously. Staff added that most large issuers spoke positively about AI’s potential.
The OSC said the findings in this study show that AI integration across different sectors is becoming more critical and that issuers are increasingly adopting a balanced perspective that weighs the technology’s opportunities and risks to their business.
“As issuers become more attuned to the potential benefits and risks of AI, we see a more nuanced discussion emerging,” Byberg said in the news release.
OSC staff study
In their analysis, staff members in the OSC’s thought leadership division considered MD&A filings from S&P/TSX composite index issuers over a decade, the OSC’s news release explained.
Staff used natural language processing (NLP) tools to analyze the sentiment of the text surrounding each reference to AI. The OSC said its staff tested new large language models (LLMs), including the one used for OpenAI’s ChatGPT, to perform analyses that usually rely on traditional NLP models.
Recent research suggests that these new LLMs do well with NLP tasks such as analyzing and summarizing text and categorizing sentiment and can potentially assist securities regulators with their research and analysis, the OSC’s staff said.
“However, our work, to date, is very much a proof of concept,” the OSC’s staff noted in the study. “We caution readers that our results, while interesting, do not assess in any way the adequacy of issuers’ disclosure with respect to AI.”