447
edits
News-April-30-2026: Difference between revisions
Add Day 4 distillation admission coverage
(Add White House/Anthropic Mythos story to daily digest) |
(Add Day 4 distillation admission coverage) |
||
| Line 10: | Line 10: | ||
Source: [https://www.nbcnews.com/tech/tech-news/elon-musk-testifies-openai-trial-sam-altman-rcna201045 NBC News], [https://www.law360.com/technology/articles/musk-accuses-openai-atty-tricking-jury Law360] | Source: [https://www.nbcnews.com/tech/tech-news/elon-musk-testifies-openai-trial-sam-altman-rcna201045 NBC News], [https://www.law360.com/technology/articles/musk-accuses-openai-atty-tricking-jury Law360] | ||
---- | |||
== Musk v. Altman: Day 4 — Distillation Admission (April 30, 2026) == | |||
On the fourth day of the ''Musk v. Altman'' bench trial, '''Elon Musk''' appeared to admit under cross-examination that his AI company '''xAI''' had used '''OpenAI's models''' to train its own through the process of '''distillation''' — a technique where one AI model is trained to mimic another. When OpenAI attorney William Savitt asked whether xAI had distilled OpenAI models, Musk replied that "generally all the AI companies" do it, and when pressed, said "partly." He characterized it as "standard practice to use other AIs to validate your AI."<ref name="wired-distill2">[https://www.wired.com/story/musk-openai-xai-distillation-trial/ Wired, "Elon Musk Seemingly Admits xAI Has Used OpenAI's Models to Train Its Own," April 30, 2026]</ref><ref name="verge-distill2">[https://www.theverge.com/ai-artificial-intelligence The Verge, "Musk admits xAI distilled OpenAI models," April 30, 2026]</ref><ref name="ars-distill2">[https://arstechnica.com/ai-artificial-intelligence/ Ars Technica, "Musk v. Altman: xAI distillation admission," April 30, 2026]</ref> | |||
The admission was notable given OpenAI's prior efforts to block distillation by foreign competitors, particularly Chinese AI labs. The Trump administration had also announced in April 2026 that it would share information with US AI companies about foreign distillation threats. Anthropic had previously blocked both OpenAI's and xAI's access to its Claude models over terms of service violations. | |||
''See also: [[Musk v Altman et al|Musk v. Altman case page]]'' | |||
---- | ---- | ||