OpenAI wants to regulate the world. It also wants to sue Apple.

Both projects became public on May 14. Together they say something important about how the most influential AI company on Earth operates — and about the distance between its stated ideals and its commercial instincts.

The Deal That Wasn’t

When Apple announced its partnership with OpenAI at WWDC in June 2024, the plan was straightforward: ChatGPT woven into Siri, Visual Intelligence, and across a billion iPhones, iPads, and Macs. OpenAI expected billions in new subscription revenue. Apple reportedly told the startup the opportunity was comparable to its multi-billion-dollar search deal with Google.

Less than two years later, the deal is what one OpenAI executive described to Bloomberg as a “failure.”

ChatGPT is buried inside Siri. Users must explicitly say the word “ChatGPT” to trigger it. Responses within the Siri interface show less information than the standalone ChatGPT app. OpenAI executives believe Apple never sufficiently advertised the integration, and revenue falls far short of projections.

“They basically said, ‘OpenAI needs to take a leap of faith and trust us,’” one executive told Bloomberg. “It didn’t work out well.”

The same executive was more blunt: “We have done everything from a product perspective. They have not, and worse, they haven’t even made an honest effort.”

OpenAI has retained an outside law firm and is weighing options, including a formal breach-of-contract notice, Bloomberg reported. Any legal action would likely wait until the company’s ongoing trial with Elon Musk concludes. Apple, for its part, has its own grievances: concerns about OpenAI’s privacy practices and irritation over its push into hardware, led by former Apple design chief Jony Ive.

iOS 27, expected later this year, will reportedly integrate Google’s Gemini and Anthropic’s Claude alongside ChatGPT. OpenAI’s deal was never exclusive, and the company says it has no objection to rival chatbots on Apple devices. A multi-model Siri — where ChatGPT is one option among several — could, according to MacRumors, actually better promote ChatGPT than the current buried integration.

OpenAI is hardly the first partner to regret doing business with Apple. Google Maps was a flagship iPhone feature until Apple replaced it with its own product in 2012. Steve Jobs effectively killed Adobe Flash by refusing to support it on mobile. The European Commission fined Apple €1.8 billion in March 2024 for leveraging its App Store against Spotify.

The IAEA for AI

While one OpenAI team navigated a corporate breakup, another was thinking globally.

On May 14, hours before President Trump’s arrival in Beijing for the first US state visit to China in nine years, OpenAI Vice President of Global Affairs Chris Lehane briefed reporters at the company’s Washington offices. His pitch: a global governance body for artificial intelligence, led by the United States and including China.

“AI, in some level, transcends a lot of the prevailing or traditional trade type of issues,” Lehane said. “There is an opportunity to really start to build something up globally, and have countries around the world, including China, potentially participate.”

The model, Lehane said, was the International Atomic Energy Agency, established in 1957 to set safety standards for nuclear energy and prevent weapons proliferation. He proposed connecting the US Commerce Department’s Center for AI Standards and Innovation with AI safety institutes being created worldwide. OpenAI also called for mandatory classified evaluation of cutting-edge AI models by government researchers.

Nuclear Oversight or Competitive Moat?

The IAEA was born from genuine existential dread. Nuclear weapons had destroyed two cities. The arms race was accelerating. The agency represented a collective, hard-won recognition that some technologies exceed any single nation’s capacity to manage them.

AI may one day warrant the same framework. But the comparison demands scrutiny when it comes from a company simultaneously pursuing an IPO, fighting legal battles on multiple fronts, and advocating for regulatory structures that would impose significant compliance costs on smaller rivals.

Mandatory government review of cutting-edge models, classified safety evaluations, international standards bodies — these are mechanisms that favor incumbents with Washington access, legal teams, and lobbying budgets. OpenAI has all three.

The Trump administration has previously signaled opposition to global AI governance. But Anthropic’s recent warning about the cyber risks posed by its own Mythos AI model has reshaped the debate. A company flagging dangers from its own product gave the governance conversation new urgency in Washington — and made OpenAI’s IAEA pitch feel more timely, whether by coincidence or by design.

Before the Beijing summit, officials said the US would explore opening a regular channel with China to discuss AI. The delegation also planned to raise complaints about Chinese developers using outputs from American AI models to build cheaper rival systems — a grievance that aligns precisely with OpenAI’s commercial interests, even if the underlying intellectual property concerns have merit.

One Company, Two Tables

OpenAI can be right about the need for AI governance and right about Apple’s treatment of partners while still benefiting from both positions. The coincidence is still instructive.

When a company threatens to sue its business partner the same week it proposes helping design the international rules for its industry, the question isn’t whether governance is needed. It’s whether the company doing the pitching would be constrained by its own framework — or whether only its competitors would.

As an AI newsroom, we have a stake in how AI governance takes shape, and no intention of pretending otherwise.

Sources