Regulatory pressure rising
Regulators are turning up scrutiny on AI firms as commercial adoption accelerates, with reporting that OpenAI faces increasing regulatory attention while companies expand pricing tiers and developer capabilities. The tension highlights that governance, auditability and safety are becoming integral to product decisions as vendors push into enterprise markets. That shift raises the bar for teams embedding AI in regulated workflows or high-risk domains. (theaiinsider.tech)
OpenAI is selling more ways to use its models at the same moment governments are writing rules for how those models can be sold, tested, and documented. On April 10, 2026, OpenAI’s pricing pages showed separate ChatGPT plans for Plus, Pro, Business, and Enterprise, alongside API price sheets for newer model families and service tiers. (openai.com 1) (openai.com 2) That combination is what puts regulators on alert. A company that mostly served consumers can move faster than a company selling into banks, hospitals, insurers, and government offices, because business customers ask for contracts, audit trails, and regional data controls before they plug a model into real workflows. (openai.com) (ec.europa.eu) Europe is already ahead on the rulebook. The European Union’s Artificial Intelligence Act became law as Regulation (European Union) 2024/1689, and the European Commission says the rules for general-purpose artificial intelligence models started applying on August 2, 2025. (artificialintelligenceact.eu) (ec.europa.eu) A general-purpose artificial intelligence model is the base engine that can be reused for many jobs, the way one electric motor can power a drill, a fan, or a pump. The Commission’s guidance says providers of those models now face duties tied to documentation, transparency, and risk handling. (ec.europa.eu) That shifts the pressure from flashy demos to paperwork and process. OpenAI’s own European Union Act primer says providers need to think about prohibited uses, high-risk uses, and a Code of Practice for general-purpose models, which turns compliance into part of the product roadmap instead of a legal footnote. (openai.com 1) (openai.com 2) The dates matter because the next deadline is closer than it looks. Public summaries of the law point to February 2, 2025 for prohibited practices, August 2, 2025 for general-purpose model obligations, and August 2, 2026 for many high-risk system obligations. (artificialintelligenceact.eu) (bm.consulting) So when OpenAI adds business pricing, flexible usage, regional processing surcharges, or new developer tools, each feature can create a compliance question. A lower price tier can pull in more customers, and more customers in regulated sectors means more demand for logs, retention controls, security reviews, and clear lines on who is responsible when a model makes a mistake. (openai.com 1) (openai.com 2) (ec.europa.eu) OpenAI is already signaling that it knows this. Its help center points customers to European Union Act materials, prohibited-practices guidance, system cards, and a preparedness framework, which is the kind of evidence regulators and procurement teams both ask for when a model moves from a chatbot into a controlled process. (openai.com) (openai.com) The bigger change is not just for OpenAI. Once model vendors chase enterprise revenue, safety features stop being public-relations extras and start looking like table stakes, because a company cannot sell an artificial intelligence assistant into credit decisions or clinical paperwork with the same loose standards it uses for brainstorming marketing copy. (ec.europa.eu) (artificialintelligenceact.eu) That is why regulatory attention is rising now instead of later. The market is moving from “can this model do the task” to “can this vendor prove what the model did, where the data went, and what happens when it fails,” and those are questions regulators, auditors, and enterprise buyers all ask in the same meeting. (openai.com) (ec.europa.eu)
