OpenAI CEO Sam Altman plans to brief the Trump administration and US lawmakers next week on the upcoming generation of artificial intelligence models, as the US works to create a process for reviewing the safety of cutting-edge AI systems. The meetings come as US officials develop a safety review framework expected to be completed within weeks, according to Chris Lehane, OpenAI’s head of global public affairs.
🔍 THE BOTTOM LINE
The timing is not accidental. The US government is weeks away from finalising a frontier-model safety review process, and OpenAI — the company most likely to be first under that microscope — wants to shape the rules before they are written. Altman’s briefings will cover a new family of models and “how they will affect work,” which is corporate-speak for: the models are coming, and we want you to hear about them from us, not from a leak.
The Framework Behind the Meetings
The US has been building toward a formal AI safety review process since the Trump administration’s executive order on frontier models. That order, signed earlier in 2026, directed agencies to develop a review framework for the most advanced AI systems — the kind that could pose national security risks if deployed irresponsibly.
Lehane told Bloomberg that Altman’s visit will focus partly on OpenAI’s new model family and their implications for the workforce. The framing matters: by emphasising economic impact alongside safety, OpenAI is positioning itself as a partner in the conversation, not a subject of it.
This is a different posture from the one OpenAI took during the Biden-era AI safety debates, when the company publicly supported voluntary commitments but resisted binding regulation. The Trump administration’s approach — a streamlined review process with a 30-day voluntary window, as reported in our coverage of the downsized AI order — is more industry-friendly, but it still creates a precedent for government oversight of frontier models.
What “Next Wave” Means
OpenAI has not disclosed what the new models are, but the context is telling. The company recently filed confidentially for an IPO, lost its chief reasoning researcher Barret Zoph to a second departure, and is racing Anthropic for enterprise customers after Anthropic overtook OpenAI in business adoption. A new model family — possibly GPT-6 or a new agentic line — would be the product launch that resets the competitive narrative.
The workforce impact framing also aligns with OpenAI’s recent push into AI agents reaching 10 million users after the ChatGPT Work launch. If the next models expand agent capabilities further, the economic-disruption conversation becomes more urgent — and OpenAI would rather lead that conversation than be ambushed by it.
The Regulatory Stakes
The US is not the only government building an AI review framework. China has already passed its first agentic AI regulation requiring humans-in-the-loop for autonomous agents. The EU’s AI Act continues to set the global benchmark for binding AI rules. And as we reported, Australia has retreated from AI regulation under Trump pressure, choosing infrastructure investment over rule-setting.
A US safety review process — even a lightweight one — would place America somewhere between the EU’s prescriptive approach and China’s reactive one. For OpenAI, being first through that process is both a risk and an advantage. A clean review sets a template that competitors must follow. A messy one creates regulatory uncertainty that could delay launches.
NZ Angle
New Zealand has no equivalent frontier-model review process. NZ companies using OpenAI APIs will get whatever the US framework produces — there is no local gate. But if the US process normalises government safety reviews for frontier models, the political pressure on NZ to adopt something similar will grow. The NZ-Australia AI regulation comparison we covered shows both countries are still at the “guidelines, not laws” stage. A functioning US review process could change that calculation.
❓ FAQ
What models will Altman brief officials on? OpenAI has not disclosed specifics. The briefing will cover a “new family of models” and their workforce impact, according to Chris Lehane, OpenAI’s head of global public affairs.
When will the US safety review framework be ready? Lehane said it is expected to be completed “in the next few weeks.” The framework has been in development since the Trump administration’s executive order on frontier models.
Does this mean the US will regulate AI like the EU? Unlikely. The Trump administration’s approach has been more industry-friendly — a 30-day voluntary review window rather than the EU’s mandatory conformity assessments. But it would be the first formal US process for reviewing frontier models before deployment.
What does this mean for OpenAI’s IPO? A functioning safety review process that OpenAI has already navigated could reassure investors that regulatory risk is manageable. It could also delay the IPO if the review surfaces concerns.
🔍 THE BOTTOM LINE
Altman’s meetings next week are not a courtesy call. They are the first real test of whether the US government’s frontier-model review process will be a genuine check on AI companies or a rubber stamp. The answer will shape how every other country — including New Zealand — approaches AI governance. And for OpenAI, getting ahead of the framework is worth more than any product launch.
