Sources: OpenAI and Anthropic are lobbying the Trump administration to make sure lagging competitors also have to comply with government reviews of their models
OpenAI and Anthropic are arch enemies whose leaders can't stand each other. But the two foes have formed an unholy alliance …
Context & Ripple Effects
OpenAI and Anthropic’s reported alignment is notable because the companies have also been depicted as fierce product and leadership rivals. Their shared position follows reported lobbying to restrict open-source AI models, broadening the policy focus from model availability to formal oversight of competing developers.
The move also extends Anthropic’s prior willingness to contest the boundaries of AI regulation, including its reported opposition to a federal limit on state AI rules. The common thread is an industry debate over which labs and models face which public constraints.
First-order effects
- OpenAI and Anthropic are jointly pressing the Trump administration for government model reviews that would apply to lagging competitors, rather than leaving scrutiny concentrated on the leading labs.
- If officials adopt the requested approach, affected rivals would face a new review requirement before or alongside deploying covered models; OpenAI and Anthropic would gain a shared channel for shaping how that requirement is defined.
Second-order effects
- Competitors would have to engage more directly with Washington on review standards, timing, and scope, while the two leading labs’ existing policy operations become more commercially consequential.
- A review regime paired with the reported push for tighter open-source rules could make the distinction between regulated frontier development and less-controlled model distribution a central competitive fault line.
Third-order effects
- If this coalition persists, AI competition may increasingly hinge on state-mediated compliance frameworks as well as model performance, rewarding firms able to secure legitimacy with policymakers.
- The direction of travel is uncertain: broadly applicable reviews could set common safety expectations, while uneven thresholds or enforcement could instead turn compliance design into a durable barrier to entry.
The trend: This is a data point in the shift toward state-mediated AI competition, where leading labs seek to help define the rules that govern rivals’ model access and deployment.