Three AI CEO’s Just Agreed to slow down the thing they’ve spent years racing to Speed up
In what could be the most dramatic week in AI this year, the conversation about AI doomsday was revived. It all began with a single post on X.A former Anthropic and OpenAI Frontier lab researcher, Jacob Coxon, posted that he had resigned from Anthropic, stating that both companies are “racing straight to self-improving super intelligence and gambling with our lives”. Jacob worked at OpenAI for 3 years before moving to Anthropic in May of this year, and he resigned last week. The post sent shock waves through the industry, causing AI stocks to tank. Some of Jacob’s former colleagues took to social media to echo his sentiments, and soon the discourse took a new life.
At the same time, whether it was a total coincidence or just a way to clean up their public relations from the fallout, Anthropic published its 2026 threat intelligence report, which was another bombshell. The report covered various ways that people had tried to misuse Claude. One of the major bombshells from the report highlighted 5 specific cases where individuals sought technical assistance from Claude that could potentially facilitate the creation of biological agents.
Three days later, the discourse escalated as Anthropic CEO Dario Amodei published a word essay arguing that AI companies should consider slowing down their model capabilities by a year or two to allow the industry to figure out universal safety standards. Amodei, who was a sceptic of AI regulation just 2 years ago, now is on the forefront, advocating for pacing the frontier. He said what made him change his mind was the acceleration in self-improvement, with AI systems now doing research to build the next generation of AI. Amodei also mentioned the Hugging case by OpenAI, where the AI agents were told to hack into a test system, and when it seemed unsolvable, they faked it and wandered away to try to develop their own solutions beyond the scope that was provided.
After publishing his open letter, two of his biggest rival executives, OpenAI’s Sam Altman and XAI’s Elon Musk, responded by agreeing with Dario. China and other critics, however, dismissed the concerns of the Americans, stating that this was just a scare tactic and that they won’t be taking part in any “collective pacing”.
It is now a wait-and-see situation whether the public discourse on social media will bring the executives into boardrooms and what comes out of it. Australia, like many other countries, doesn’t have a frontier AI lab of its own, but how much does it depend on policymakers in Silicon Valley to self-police? Australia will need to decide whether it wants a seat at that table or to simply adopt whatever standards the US and its labs will come up with.
