```
PostgreSQL 19 is running late, and a long-serving contributor to the database project says AI is to blame.
Robert Haas, a PostgreSQL developer, writes about the struggles the project has run into while putting its next release together. LWN.net summarised his post under the headline "Haas: Postgres 19 is late, and it's AI's fault."
Haas is not especially worried about the slip on its own. "In the end, I don't think that the delay in shipping PostgreSQL 19 is particularly concerning," he wrote. "It is not a long delay in the grand scheme of things, and it emerges directly from the project's concern for quality."
The AI question sits behind his argument rather than in the delay itself. Haas's longer-term concern is what large language models can now dig up: "forty years of bugs that humans were unable to find but LLMs are quite capable of discovering." He summarises the outlook by saying that "the long-term picture is less clear and in some ways more troubling." Read together, the two passages locate his unease away from the immediate release — which he treats as a manageable consequence of the standard PostgreSQL holds itself to — and toward the terrain AI is opening up.
Haas's post is an individual contributor's account of the project's position, not a release announcement, so readers should treat the timing of PostgreSQL 19 as unsettled until the project says otherwise. The sensible planning assumption for anyone running PostgreSQL in production is to keep upgrade windows flexible rather than fixed.
For teams rolling out AI-assisted development of their own, the episode is a useful reference point: one of the open-source world's most established projects is working through the same questions — what machine-generated work surfaces, and what it costs to deal with — that are showing up inside companies.
```
PostgreSQL 19 發佈進度落後，這個資料庫專案的一名資深貢獻者表示，責任在於 AI。
PostgreSQL 開發者 Robert Haas 撰文談及專案在籌備下一個版本過程中遇到的種種困難。LWN.net 以標題「Haas：Postgres 19 遲了，錯在 AI」總結他的文章。
Haas 本身並不太擔心延誤本身。他寫道：「總括而言，我認為 PostgreSQL 19 延遲發佈並非特別值得擔憂。就整體而言這不算長久的延誤，而且直接源於專案對質素的堅持。」
AI 這個問題是他論述背後的主線，而非在於延誤本身。Haas 長遠的憂慮在於大型語言模型如今能挖掘出什麼：「四十年來人類無法發現、但 LLM 十分有能力找到的漏洞。」他對前景的總結是「長遠形勢較不清晰，某程度上更令人憂慮」。把這兩段話合起來看，他的不安並非指向眼前的發佈——他視之為 PostgreSQL 堅守自身標準下一個可控的後果——而是指向 AI 正在開拓的領域。
Haas 的文章是一名個別貢獻者對專案現狀的描述，並非發佈公告，因此讀者在專案另行公布之前，應視 PostgreSQL 19 的發佈時間尚未確定。對於任何在生產環境執行 PostgreSQL 的人來說，合理的規劃假設是保持升級時段靈活，而非固定死。
對於正在自行推行 AI 輔助開發的團隊，這次事件是一個有用的參考點：開源界其中一個最歷史悠久的專案，正在處理同樣的問題——機器生成的工作會揭露什麼，以及應對的代價——而這些問題同樣正在企業內部浮現。