The Elon Musk-OpenAI trial is bringing out many interesting tidbits from the early days of OpenAI.
In the ongoing Musk v. Altman proceedings, OpenAI has sought to introduce a peculiar piece of evidence: a physical trophy — specifically, the rear half of a donkey — gifted to Josh Achiam, OpenAI’s chief futurist. The inscription reads: “Never stop being a jackass for safety.”
The backstory is this: when Musk departed OpenAI, he held an all-hands meeting with employees. During his remarks, Achiam pushed back, arguing that Musk was prioritising the race to beat Google over a critical safety concern. Musk, unimpressed, called him a jackass. At the very next all-hands, OpenAI employees responded by honouring Achiam with the trophy. On the stand, Achiam testified that it was none other than Dario Amodei and David Luan who presented it to him — and that Musk’s remark “was not friendly,” that he felt Musk was simply reacting to being challenged.
OpenAI is using the trophy as evidence to argue that, despite his public positioning, Musk was never as committed to AI safety as he claims. The exhibit is designed to show that when safety concerns collided with his competitive instincts, Musk chose the latter.
The irony of that history isn’t lost on anyone watching the AI space today. Dario Amodei — the man who once handed out a trophy for standing up to Elon Musk on safety grounds — is now doing business with Musk directly. Anthropic recently signed a landmark deal with SpaceX to take over the full compute capacity of Colossus 1, Musk’s Memphis data centre housing over 220,000 NVIDIA GPUs. The deal came after Anthropic’s infrastructure buckled under demand that grew 80x year-on-year — a rate CEO Amodei himself called “just crazy.”
What makes the partnership especially striking is how recently the hostility was on full display. Musk had called Anthropic “evil” and accused it of hating “Western civilisation” just months before the deal was announced. Amodei, for his part, had previously questioned whether signing massive compute agreements was financially reckless — before Anthropic’s own explosive growth forced his hand.
The computing alliance is driven by cold pragmatism on both sides. Anthropic needed GPUs immediately. SpaceX had a massive data centre running underutilised, with a public offering on the horizon that would benefit from a blue-chip customer like Anthropic on its books. As one industry observer put it: “Elon’s enemy is Sam. Dario’s enemy is Sam. Enemy of my enemy is a compute partner.”
The Amodei-Musk partnership would have seemed outlandish even six months ago. Now it is simply business. That is how fast things move in the AI industry — and the Musk-OpenAI trial, with its jackass trophies and safety debates from a decade ago, is a reminder of just how much ground has been covered since those early, scrappier days.
The two men who once represented opposing poles of the AI safety debate are now, at least commercially, on the same side. The trophy presumably still sits on Achiam’s shelf.