WASHINGTON — OpenAI announced on Thursday, 29 September 2026, that it would suspend the commercial rollout of its next‑generation language model, GPT‑6.1 Astra, citing unresolved safety concerns. The decision was disclosed in a brief statement to the press and confirmed by multiple news agencies.
Company cites internal testing
OpenAI said that a series of internal evaluations revealed the model could be exploited to produce disinformation, automate phishing attacks, and generate content that evades existing moderation tools. The lab described the findings as “significant enough to warrant a pause while additional safeguards are developed.” The statement did not disclose the specific technical issues or the timeline for a revised release.
Model capabilities and expectations
GPT‑6.1 Astra was positioned as a major upgrade over the earlier GPT‑6 series, promising higher reasoning ability, multimodal input handling, and faster response times. Industry analysts had expected the model to be integrated into a range of enterprise products, from customer‑service chatbots to code‑generation assistants. The abrupt halt has prompted clients to reassess short‑term AI roadmaps.
Regulatory environment intensifies
The move comes amid heightened scrutiny of advanced AI systems worldwide. The European Union’s AI Act, which entered full force in 2025, imposes strict conformity assessments on high‑risk models. In the United States, the White House’s AI Bill of Rights, released in early 2024, urges developers to embed robust risk‑mitigation measures before public deployment. OpenAI’s pause aligns with a broader industry trend of pre‑emptive safety reviews.
Industry reaction
Several technology firms issued cautious statements. A spokesperson for Microsoft, a long‑time partner of OpenAI, said the company remains “committed to responsible AI development” and will work with OpenAI to understand the safety gaps. Alphabet’s DeepMind declined to comment beyond noting its own internal safety protocols.
Investor and market impact
OpenAI’s parent company, OpenAI LP, is privately held, but the news was noted by venture‑capital investors who have previously backed the lab. A senior partner at Andreessen Horowitz, who declined to be named, described the pause as “a prudent step that underscores the maturity of the organization’s risk management culture.”
Future roadmap
OpenAI said it will convene an internal task force to address the identified risks and will publish a revised safety framework before any future release. The lab also indicated that it will engage external auditors and seek input from the Partnership on AI, a multi‑stakeholder consortium.
