Claim Assessed as Largely Accurate

The assertion that OpenAI shelved the launch of its next-generation AI model, known as Astra, over safety concerns was reviewed against primary sources. The assessment drew on a set of 12 primary sources and concluded that the claim stands, with the finding recorded in Korean as '대체로 사실' (mostly true).

Uncertainty Over Details Remains

The verification confirmed the core of the report — that the model's release was scrapped and that safety concerns were the stated reason. Details beyond this central claim, including the model's intended capabilities and timeline, remain unclear based on the material reviewed.

OpenAI's decision to withhold Astra underscores the growing weight the company places on safety evaluation ahead of next-generation model releases.

Verdict: Mostly True