OpenAI previews powerful new ‘Astra’ model for U.S. officials amid policy battles

According to recent U.S. reporting, OpenAI CEO Sam Altman has been quietly briefing White House and congressional officials on the company’s next‑generation AI model, codenamed “Astra.” The system is described as OpenAI’s most powerful model to date, expected to surpass the capabilities of the current ChatGPT lineup and to play a central role in upcoming policy debates.

Washington is in the middle of building a federal framework for AI, with questions around safety evaluations for large models, transparency obligations, and conflicts with stricter state‑level rules. By giving policymakers an early look at Astra’s technical profile and safety measures, OpenAI appears to be trying to shape the conversation and head off regulations it sees as overly restrictive.

Regulators and lawmakers, however, remain wary. Past frontier models have been linked to issues like large‑scale misinformation and unpredictable behavior by autonomous AI agents, and some are calling for special oversight for systems on Astra’s scale. Even before its public release, the model is becoming a flashpoint in the broader struggle to define “the rules of the game” for the next wave of AI.