The Ontario Securities Commission recently released staff research that analyzes the frequency and sentiment of Canadian listed issuers’ references to Artificial Intelligence (AI) in their financial disclosures.

Here’s how the OSC’s news release summed it up:

• OSC staff analyzed the Management Discussion and Analysis (MD&A) filings from S&P/TSX Composite Index issuers over a 10-year period.  The exploratory study also provided OSC staff with an opportunity to test new large language models (LLMs) for sentiment analysis.
• The exploratory research revealed three key findings:

1. Increased AI Mentions: The OSC observed a substantial rise in the number of issuers mentioning AI in their MD&As, reflecting the growing importance and integration of AI technologies across various sectors.
2. Shift in Tone: The 2024 filing year marks a turning point in the overall tone used by issuers to discuss AI. Previously most issuers that discussed AI had an overall positive sentiment, however along with more issuers mentioning AI, there has been a substantial increase in negative sentiment, indicating a more balanced view of AI’s risks and opportunities.
3. Diverging Sentiments: Issuers in the Finance and Information sectors exhibit a distinct divergence in sentiment compared to those in other industries. While Finance and Information sectors maintain a more positive outlook on AI, other industries are increasingly cautious.

As indicated above, one of the more interesting aspects of the exercise is that it itself made use of AI:

• we asked OpenAI’s GPT-4o mini  – one of the LLMs that powers ChatGPT – to classify the text surrounding issuers’ mentions of AI as either positive, negative or neutral. Model classification is not perfect, but GPT-4o mini performed better than the other, more traditional, sentiment analysis models that we tested: GPT-4o mini’s sentiment classification matched our researchers’ classification for around 80% of the text in our test sample…Another caveat is that a model can only analyze the text provided to it, potentially missing useful context included in other parts of a filing.

The document indicates that the OSC and other regulators “are exploring how AI tools can improve regulatory work, for example, by supporting enforcement investigations, facilitating disclosure reviews, and complementing research.” And of course it would be foolish for them not to do that. There’s a major disconnect though between such progressive innovation and a stuck-in-the-past mindset which neurotically fusses over the precise content of old-fashioned linear disclosures. For example, as we addressed here (and perhaps contributing somewhat to the more balanced disclosures noted above), the Canadian Securities Administrators had addressed this area in the most recent report on its Continuous Disclosure Review Program:

• Issuers adopting new technology will need to consider whether disclosures regarding the use of and variety of evolving risks associated with new technology is necessary. For example, if an issuer began using artificial intelligence (AI) systems in its product or service offerings, the issuer should only disclose the use of AI systems with a reasonable basis for doing so, otherwise such disclosure would be overly promotional… Issuers should consider their disclosure obligations regarding an issuer’s reliance on AI and the potential exposure to material risks on an entity-specific basis. We expect investors would generally consider the following information material: the source and providers of the data that each AI system uses in order to perform its functions, whether the AI system used by the issuer is being developed by the issuer or supplied by a third party, the impact that the use, development or dependency on AI systems is likely to have on the issuer’s business, results of operations and financial condition, whether there have been any incidents where AI system use has raised any regulatory, ethical or legal issues, and any other concerns that arose with the adoption of AI systems. Balanced disclosure that includes a discussion of the benefits and risks to using AI systems should be provided…

A question might arise over whether the OSC’s citing of AI as a contributor to its recent research isn’t itself “overly promotional,” given the modest import of those findings. But anyway, the report indicates that the content of the negative mentions “generally referred to risks which could range from, for example, considerations about the risk of competitors adopting AI at a faster pace, to concerns around AI’s complexity, security risks and whether tools would function as intended.” The implication there seems to be that companies feel themselves getting pulled into an AI-shaped world, whether they want to or not, with regulators accordingly amending their approach and focus. Having said that, past innovations such as XBRL amounted to much less than anticipated, and it seemed not long ago that we were entering a new era of mandated sustainability disclosure, whereas now, well, it doesn’t. Perhaps AI will be different, and the regulators will get pulled further than they can currently envisage…

The opinions expressed are solely those of the author.