A new federal lawsuit is accusing four of the biggest companies in artificial intelligence of crossing the line from discussing AI safety into illegally coordinating with one another to slow the development of their competing technologies.
The complaint targets Anthropic, OpenAI, SpaceXAI and Google and was filed Sept. 18 in the U.S. District Court for the Northern District of California. The plaintiffs contend that the companies violated federal antitrust law by agreeing to collectively reduce the pace of AI development instead of making those decisions independently.
At the center of the case is a growing debate within the AI industry over whether the race to build increasingly capable systems is moving faster than companies can understand or control the risks those systems may create.
The lawsuit argues that the alleged coordination became particularly visible on Sept. 12, when Anthropic CEO Dario Amodei published an essay calling for major AI developers to cooperate on slowing the pace of advancement while stronger safety protections are put in place.
OpenAI CEO Sam Altman, SpaceXAI CEO Elon Musk and Google DeepMind co-founder and chair Demis Hassabis each publicly responded to Amodei’s proposal that same day in ways that supported greater coordination around AI safety and development speed, according to the complaint.
The plaintiffs maintain that the public exchange amounted to more than several executives independently expressing similar concerns. They allege the companies were effectively agreeing that development across the industry should proceed more slowly than it otherwise would in a competitive market.
The lawsuit says the groundwork for that cooperation may have begun months earlier.
It points to a statement signed in July 2026 by senior employees from several leading AI laboratories. That statement acknowledged that companies face intense competitive pressure that makes it difficult for any one laboratory to slow development on its own while rivals continue moving forward.
The July statement called for government involvement in creating a broader global effort to reduce the speed of increasingly automated AI development.
That competitive pressure has become one of the central problems facing companies that say they are concerned about advanced AI safety. A company that voluntarily slows development could potentially lose customers, investment and technological advantages to competitors that continue releasing more powerful systems.
The lawsuit argues, however, that antitrust law does not allow competing companies to solve that problem by privately agreeing with one another to restrain development.
Four named plaintiffs brought the case. Each pays for access to one of the major commercial AI services involved: OpenAI’s ChatGPT, Anthropic’s Claude, SpaceXAI’s Grok or Google’s Gemini.
Their attorneys are seeking to represent a nationwide class consisting of other consumers who pay for subscriptions to those AI services.
The plaintiffs argue that subscribers are harmed if competing companies collectively slow innovation because consumers are paying for services with the expectation that those products will continue improving through normal market competition.
In their view, an industrywide agreement to reduce the speed of development could result in subscribers receiving less technological progress and therefore less value from their paid subscriptions than they would receive if the companies continued competing independently.
The plaintiffs are not arguing that individual AI companies must develop their systems as quickly as possible.
The complaint explicitly allows for the possibility that Anthropic, OpenAI, Google or SpaceXAI could independently conclude that safety concerns justify delaying a model, reducing capabilities or slowing development.
What the plaintiffs object to is competitors making that decision together.
Their attorneys argue that competition requires every company to take responsibility for its own safety decisions rather than replacing individual judgment with a collective restraint agreed upon by major competitors.
Lead attorney Nick Rowley has also framed the dispute as involving more than subscription prices or product features. He argues that decisions surrounding the safety of potentially powerful AI systems should not be controlled primarily through private agreements among some of the world’s largest for-profit technology companies.
None of the four companies had publicly responded to requests for comment on the lawsuit as of Saturday, according to reporting on the case.
Amodei had already acknowledged that coordinated discussions among competing AI companies could create antitrust concerns before the lawsuit was filed.
In his essay, he suggested the federal government could help facilitate or authorize discussions among competing laboratories about safety.
Amodei said the government would not necessarily need to participate directly in every conversation but suggested that a narrow antitrust exemption could be created for certain types of AI safety discussions.
Such an exemption could potentially allow companies that would normally be prohibited from coordinating business decisions to discuss specific safety-related issues without automatically creating antitrust liability.
Altman took a somewhat different position in his public response.
The OpenAI CEO said his company supported the idea of a federal framework establishing consistent AI safety requirements. At the same time, he argued that companies did not have to wait for Congress to pass legislation or for an antitrust exemption before taking steps intended to give the public greater confidence in AI safety.
The discussion comes as AI companies increasingly debate the possibility that advanced systems could eventually operate with less human oversight, behave unpredictably or become capable of carrying out complex tasks that their creators struggle to monitor.
Industry leaders have discussed shared safety standards for years, and several companies have participated in voluntary initiatives intended to establish common testing, security and risk-management practices.
The plaintiffs say they are not opposed to those companies asking Congress, the White House or federal agencies to impose AI regulations.
They also say they do not object to the companies formally asking the government for an antitrust exemption that would legally permit certain types of cooperation.
Instead, the lawsuit challenges what the plaintiffs characterize as private coordination occurring without that government authorization.
Whether the industry could obtain such an exemption is uncertain.
President Donald Trump has publicly rejected recent calls for stronger AI restrictions and has characterized efforts to limit development as a threat to American technological leadership.
Trump has questioned why major technology companies would advocate regulations that could restrict their own businesses. On Saturday, he also announced plans to establish an AI task force and appoint an AI czar, although he provided few details about how either would operate.
The administration has repeatedly emphasized the importance of American AI companies maintaining an advantage over China, making proposals that could deliberately slow U.S. development politically contentious.
Several Democratic leaders and candidates have called for broader federal action on AI regulation, while many Republicans have taken positions closer to Trump’s argument that excessive restrictions could weaken American companies in the international AI race.
Republican Sen. Josh Hawley of Missouri has also expressed strong opposition to giving the largest AI companies an exemption from antitrust law.
During a recent Senate hearing, Hawley argued that he would not support allowing some of the world’s most powerful corporations to coordinate under an antitrust exemption because doing so could give them an opportunity to suppress competition.
That concern illustrates one of the contradictions now confronting policymakers.
Some AI researchers and executives argue that companies cannot safely slow development unless competitors agree to do the same, because unilateral restraint could cause a company to fall behind.
Antitrust law, however, is specifically designed to prevent competitors from privately coordinating actions that reduce competition.
The new lawsuit could force courts to examine where legitimate cooperation on safety ends and unlawful coordination among competitors begins.
It could also influence how AI companies communicate with one another about future safety concerns. If private coordination creates antitrust exposure, companies may increasingly seek formal government supervision or legal exemptions before entering into industrywide agreements.
For paid users of ChatGPT, Claude, Grok and Gemini, the plaintiffs are asking the court to treat the issue as a consumer competition case as well.
Their argument is that subscribers should receive the benefits produced when competing companies independently attempt to build better products, and that major rivals should not be allowed to collectively determine how quickly those products improve.
The companies, meanwhile, have been confronting growing warnings that the rapid development of increasingly autonomous AI systems could eventually create risks that ordinary commercial competition is poorly equipped to address.
The federal court will now have to consider whether the alleged cooperation represents a lawful attempt to address those risks or an anticompetitive agreement among some of the most powerful companies in the rapidly expanding artificial intelligence industry.
Source: ABC15 Arizona



