Elon Musk testifies that xAI trained Grok on OpenAI models

OpenAI and Anthropic person been connected the warpath lately against third-party efforts to bid caller AI models by prompting their publicly-accessible chatbots and APIs, a process known arsenic “distillation.”

That speech has focused connected Chinese firms using distillation to make open-weight models that are astir arsenic susceptible arsenic U.S. offerings, but disposable astatine a overmuch little cost. However, tech workers person wide assumed that American labs usage these techniques connected each different to debar falling down competitors.

Now, we cognize it’s existent successful astatine slightest 1 case: connected the basal in a California national court connected Thursday, Elon Musk was asked if xAI has utilized distillation techniques connected OpenAI models to bid Grok, and helium asserted it was a wide signifier among AI companies. Asked if that meant “yes,” helium said, “Partly.”

Musk is successful the process of suing OpenAI, CEO Sam Altman, and Greg Brockman, alleging they breached the archetypal nonprofit ngo for OpenAI by shifting the entity to a for-profit structure. That proceedings began this week, featuring grounds from the tech leader.

Musk’s admittance is notable due to the fact that distillation threatens AI giants by undermining the vantage they’ve built by investing successful compute infrastructure. This allows different bundle makers to make models that are astir arsenic susceptible connected the cheap. There’s nary tiny magnitude of irony here, fixed the bending and alleged breaking of copyright rules by frontier labs successful their hunt for capable information to bid their models.

It’s nary astonishment that Musk’s xAI, which started successful 2023, years aft OpenAI, would effort to larn from the then-leader successful the field. It’s not wide that distillation is explicitly illegal, but alternatively whitethorn interruption the presumption of work companies acceptable for the idiosyncratic of their products.

OpenAI, Anthropic, and Google person reportedly launched an inaugural done the Frontier Model Forum to stock accusation astir however to combat distillation attempts from China. These typically impact systematic querying of models to recognize their interior workings. To halt the efforts, frontier labs are moving to forestall users from making suspicious wide queries.

OpenAI did not respond to a petition for remark connected Musk’s admittance astatine property time.

Later successful his testimony, Musk was asked astir a assertion helium made past summertime that xAI would soon beryllium acold beyond immoderate institution too Google. In response, helium ranked the world’s starring AI providers, saying Anthropic held the apical spot, followed by OpenAI, Google, and Chinese unfastened root models. He characterized xAI arsenic a overmuch smaller institution with conscionable a fewer 100 employees.