AI Press Releases: Data Reveals Which AI Platforms Read the Most TMX Newsfile Releases
- Melissa Strle
- 6 hours ago
- 5 min read
Part of TMX Newsfile’s ongoing look at how AI is reshaping corporate disclosure and communication.
According to TMX Newsfile data, OpenAI and Microsoft now account for more than 50% of all AI activity detected reading TMX Newsfile press releases, showing how deeply these systems engage with corporate news. Each time a company issues a press release, it now reaches more than journalists and investors. It is also being read and analyzed by AI models such as ChatGPT, Copilot, and Grok, which use corporate disclosures to train their systems and generate accurate, real-time answers for users.
This shift is redefining corporate communication. Your AI press releases no longer just inform people. They teach machines and this shapes how AI represents your company in future answers and summaries.
In this article:
Meet Your New Top Readers of AI Press Releases
Our analysis of AI-driven activity across the TMX Newsfile network shows that in 2025, OpenAI (ChatGPT) and Microsoft (Copilot) are the top readers of corporate press releases, together representing over half of all AI activity.
OpenAI (ChatGPT) - Represents about 30% of all AI-driven press release impressions. OpenAI uses press releases both for training (model learning) and grounding (real-time factual reference), giving it a strong footprint across both types of AI activity.
Microsoft (Copilot) - Accounts for roughly 25%, driven by grounding activity across Bing and Microsoft 365, where Copilot retrieves and summarizes live factual content from sources like press releases to support user queries.
ByteDance (Doubao), X (Grok), and Amazon (Titan/Alexa) - Together represent about 35%, with activity primarily linked to training their respective large language models.
Together, these interactions come from a small group of powerful AI systems. Here’s a quick look at which companies operate each model.
This data confirms that your company’s news is being read and processed by the most influential AI systems shaping how information is created, summarized, and shared.
Two Ways AI Reads Your News: The Student and The Fact Checker
AI systems read press releases in two distinct ways, and understanding both explains why your company’s content continues to matter long after publication.
The Student (Training)
Models such as Doubao, Claude, LLaMA, and Titan/Alexa read archived releases to learn how businesses communicate. They study how earnings are written, how CEOs announce change, and how information is structured.
Every clear, factual release helps teach AI what credible corporate communication looks like and becomes part of its long-term memory.
The Fact Checker (Grounding)
Systems such as ChatGPT, Copilot, Gemini, and Perplexity use press releases in real time to verify information and answer questions.
When someone asks about your company’s latest results, these AIs pull directly from your releases to provide accurate, up-to-date responses.
Together, these two modes—learning and referencing—show why strong, factual press releases now shape both how AI learns and how it reasons.
As AI systems continue to read and respond to company news, the next step is making sure your releases are written to be understood by them. This is where Answer Engine Optimization (AEO) comes in.
From Crawling to Context: How AI Reading Habits Have Evolved
When we looked back at AI-driven activity from 2022 through 2025, one company initially stood far ahead: ByteDance (Doubao). During this early period, Doubao led all AI systems and accounted for roughly 30 percent of all AI-driven press release interactions across our network.
This dominance reflected the first phase of AI development, which focused heavily on large-scale training. Doubao’s crawlers scanned billions of words, including press releases, to teach its models how businesses communicate, report financials, and launch products.
In 2025, the picture looks very different. The landscape has shifted toward a more balanced model where both training and grounding play vital roles.
OpenAI (ChatGPT) has grown significantly in training-based activity, while Microsoft (Copilot) continues to lead in grounding. Together, they now account for just over 50% of all AI-driven interactions. Training remains a powerful force, but grounding is rapidly becoming an equally important layer.
This evolution shows that AI reading behavior is no longer one-dimensional. Press releases are being used both to train tomorrow’s models and to inform today’s real-time answers. In doing so, they have become one of the most valuable content sources for both learning and grounding.
Inside ByteDance’s Doubao: The Model Behind the Numbers
One AI model in particular helps illustrate how training activity works.
ByteDance’s Doubao has been one of the most active crawlers on the TMX Newsfile network, using press releases as structured, factual material to teach its language models how businesses communicate and announce new developments.
Doubao’s training activity demonstrates why press releases are such valuable content for large language models. Their consistent format, credible information, and timely updates make them ideal material for AI systems that are learning how to write, summarize, and interpret corporate news.
AI Press Releases by the Numbers: 2022 to 2025
The data below shows how AI readership of TMX Newsfile press releases has shifted from the “training boom” years to today’s stage.
Comparing long-term trends to 2025 YTD reveals which AI models are still in heavy training cycles and which have slowed. There has been a clear shift in leadership from ByteDance to OpenAI.
Overall, ByteDance’s intensive data collection has eased, while OpenAI is now driving the next major training wave. Training remains the dominant behavior, and the platforms leading it are the ones shaping how AI learns from real-world corporate news.
But, real-time grounding is also at the forefront and is expected to increase over the coming years.
Why AI Press Releases Matter for Communicators
AI is no longer a passive technology. It’s an active consumer of your news. Every press release now speaks to two audiences:
Humans — investors, journalists, and stakeholders.
Machines — AI models that learn from, ground in, and quote your words.
The takeaway: AI press releases are your brand’s data handshake with the future. The clearer, more structured, and more factual your communication, the more accurately AI systems will represent your company, in answers, summaries, and search results.
This blog revealed which AI platforms are most actively reading TMX Newsfile press releases — with OpenAI and Microsoft now leading by a wide margin. We also saw how AI systems use press releases in two ways: to train their models and to ground real-time answers. Together, these insights show that press releases now play a critical role in teaching AI how to understand and communicate corporate information.
Press releases have always shaped market perception. Now they shape machine perception, too. The way you write your press releases today will determine how AI describes your company tomorrow.
Next, we’ll explore how fast this AI audience is growing — and how it compares to traditional human readership. The results might surprise you.
