OpenAI used YouTube data to train some of its models: Report
The Hindu
OpenAI, the company behind the AI-powered chatbot ChatGPT, used YouTube data to train some of its AI models.
OpenAI, the company behind the AI-powered chatbot ChatGPT, used YouTube data to train some of its AI models, reported tech outlet The Information, citing an anonymous source.
The outlet also reported that Google, which owns YouTube, has been using the video sharing platform’s data to train its own model Gemini.
As more Big Tech companies pivot to developing their AI capabilities or AI-powered offerings, there have been debates about the scraping of data, including copyrighted media, for the purpose of training models.
While companies behind text-to-image generators have been subject to lawsuits revolving around violating the copyright of artists, many large language models are being developed in secrecy with little to no transparency about the content in their training data.
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
In April, billionaire Elon Musk threatened to sue Microsoft, which has invested heavily in OpenAI. Musk alleged that the software maker “trained illegally with the use of Twitter data.”
“We are judges and therefore, cannot act like Mughals of a bygone era ... the writ courts in the guise of doing justice cannot transcend the barriers of law,” the High Court of Karnataka observed while setting aside an order of a single judge, who in 2016 had extended the lease of a public premises allotted to a physically challenged person to 20 years contrary to 12-year period stipulated in the law.
The High Court of Karnataka on Monday declined to interfere, at present, in the investigation against a Bharatiya Janata Party worker, who is among the accused persons facing charges of circulating obscene clips, related to “morphed” images and videos clips related to Prajwal Revanna, former Hassan MP, in public domain through pen drives and other modes.