Court Filings Cited in Claim Over OpenAI's Use of News Articles Fall Short of Full Support
A claim circulating in Korean media and technology circles asserts that a publicly disclosed court document shows OpenAI used some 10 million news articles without authorization. The tech desk examined the claim against 10 primary sources. The record reviewed confirms that court materials touching on OpenAI's use of press articles have entered the public domain, but the same record does not fully substantiate the claim as written โ leaving its headline figure and its legal characterization unproven on the evidence examined.
What Is Being Claimed
The claim rests on three elements: that a court document has been disclosed, that the document shows OpenAI used approximately 10 million articles from news publishers, and that this use was unauthorized. The assertion has drawn attention because it speaks directly to the central dispute in generative AI โ whether LLM developers may train on copyrighted news content without a license, a question with sharp consequences for publishers in Korea and abroad.
What the Record Supports
์ถ์ฒ: ์กฐ์ฌ ์ถ์ฒ
The 10 primary sources reviewed do corroborate the core of the disclosure: court filings referencing OpenAI's use of news publisher content have been made public. To that extent, the claim is anchored in documented material rather than rumor, and readers can verify the existence of such filings in the court record.
Where the Claim Overreaches
The reviewed materials do not establish the full assertion. In particular, the sources examined do not show that the disclosed documents themselves confirm unauthorized use at the scale of 10 million articles. The number and the legal framing โ the two details that give the claim its force โ are not borne out as stated by the evidence on hand. Until a court makes findings on the scale and the legality of the use in question, the claim should be treated as ahead of the documentary record.
Bottom Line
Court materials concerning OpenAI's use of news articles have genuinely been disclosed, but the fuller assertion โ that those documents prove unauthorized use of 10 million articles โ is not fully supported by the sources reviewed, making the claim Partly True.