OpenAI's global policy chief Chris Lehane confirmed Tuesday that the company has been engaging with rivals Anthropic and Google DeepMind on AI safety for several weeks. Lehane argued that no antitrust waiver is necessary for the three AI firms to coordinate on safety matters. U.S. antitrust regulators are already signaling their skepticism toward giving AI companies an exemption to coordinate on safety; Federal Trade Commission Chair Andrew Ferguson said Tuesday he would be deeply suspicious of such requests. Ferguson added that he sees the AI companies as asking for barriers to entry that will insulate their business from challengers, and called it moat digging. That is the entire problem. The vendors say the coordination is legal. The regulator says it looks like collusion. The APIs you call today do not care which position wins.
I run production systems on OpenAI, Anthropic, and Google. The construction ERP at coenconstruction.com calls GPT-4 for document extraction. The estimating system at estimate.pro calls Claude for bid analysis. The punch-list dispute system at thepunchlist.ai calls Gemini for timeline reconstruction. The vendors coordinate on safety. They do not coordinate on uptime, API versioning, or the notice period before a model gets deprecated. Anthropic CEO Dario Amodei published a 3,800-word essay Saturday urging restraining development of the most advanced systems so researchers can better understand potential threats. OpenAI CEO Sam Altman and SpaceXAI chief Elon Musk quickly embraced the missive, with Altman saying OpenAI would join Anthropic in embedding third-party evaluators into the company to monitor for safety. The safety coordination is public. The API stability guarantee is not.
The talks have been running for weeks. The waiver has not.
On Tuesday Lehane admitted the three companies have been in talks for several weeks regarding AI safety. Some, including Altman, have noted that such talks could put the companies at risk of violating antitrust law if the coordination is found to suppress competition. Amodei's essay proposed a narrow government waiver allowing such safety coordination, but Lehane reportedly said the firms don't need one. Lehane said the companies' coordination on AI safety has ample precedent, citing the airline industry as an example, and noted there are many instances over time where companies are trying to help each other on safety. The airline analogy is useful. When airlines coordinate on safety they do not coordinate on ticket prices, route networks, or which planes get retired. When AI labs coordinate on safety they are also coordinating on the models that power every API I call. The coordination may be legal. The surface area is larger.
The CNBC report confirms the talks are real. The Bloomberg report confirms OpenAI believes no waiver is required. The FTC response confirms the regulator is watching. Reuters separately noted that the Federal Trade Commission has expressed suspicion toward calls for AI antitrust exemptions, a signal that regulators are watching how these agreements are structured. I do not have a legal opinion on whether the coordination violates antitrust law. I have an operational opinion on whether a vendor embroiled in an antitrust dispute is a stable dependency. It is not. The dispute does not need to be settled for the vendor to change behavior, pull a model, or reprice an API to demonstrate independence. The coordination changes the risk profile even if it never changes the law.
The safety standards will not ship with an SLA.
Lehane said OpenAI supports a provision in the FRONTIER Act that would force top frontier labs to allow independent verification organizations into their companies to ensure models are developed safely. Any AI bill becoming law before the end of 2026 is a long shot; the House departs Washington on Thursday and has no scheduled sessions before voters head to the polls in November. The bill will not pass this year. The coordination is happening now. The legal framework is a proposal. The API I call today does not wait for the framework to finalize before it processes a request. The vendors are coordinating on standards that do not yet have legislative backing, regulatory clarity, or a finalized scope. That is an unstable foundation for a production dependency.
I have written before about vendors that draw red lines the judge says are legal. A vendor that coordinates with competitors on safety and insists no waiver is needed is drawing a different kind of line. The line may hold. It may not. The FTC chair has already called the coordination deeply suspicious. The production system does not care whether the vendor's legal theory is correct. It cares whether the vendor is stable enough to run unchanged for the next six months. A vendor in active coordination talks with its two largest competitors, operating without a waiver the CEO of one competitor says they need, under skeptical regulatory scrutiny, is not that vendor. The API works today. The legal position may not work tomorrow. The gap between those two facts is the entire risk.
The labs say they are coordinating on safety, not price. The regulator hears moat digging. The API does not distinguish between the two.
The talks will continue. The waiver will not arrive. The FTC will keep watching. I will keep calling the APIs, because there is no alternative that does not also carry vendor risk, and because the systems that depend on these models are already in production. But I will not pretend the coordination is free. It is a signal that the vendors believe they need to work together to manage risk. That makes them more aligned. It also makes them more vulnerable to regulatory action, reputational pressure, and the operational chaos that follows when three competitors try to set standards without a legal safe harbor. The API shipped. The safety framework did not.