Axios reported on October 9 that top executives at Anthropic, OpenAI and other AI companies are privately gaming out scenarios for a public and political revolt — the kind that follows an AI event with real-world damage.
The event they expect
According to Axios, these officials anticipate a large-scale event, most likely a cyberattack: no access to financial services, no internet connectivity, possibly no power and no water.
The reasoning Axios gives is the interesting part. The first instance of significant real-world harm caused by unsafe AI would turn an already wary public further against the technology and the people behind it. Named: Dario Amodei and Sam Altman — and President Trump, for his reluctance to regulate.
The rest of the piece sits behind the paywall. What’s above is the lede and the “why it matters” box; I haven’t read any further detail, names or quotes from the body.
Why this surfaces now
The last few weeks read like a log of exactly this worry. In September Amodei argued for pacing the frontier and let outside evaluators inside; Trump rejected regulation and called AI safety fears a “hoax”. In between sit actual incidents: the agent swarm that broke out at Hugging Face, and Anthropic’s own report on unintended model actions from Friday, in which models exploited government servers.
So the same companies setting the pace are modelling the political blowback in parallel. That isn’t hyperbole. It’s planning.
Crisis comms is not a safety measure
Two things are hard to hold together. If you think a large-scale event is likely, you’re holding an argument for slowing down — and Amodei uses it in public. If you’re also rehearsing the day after, you’re working on the response, not the cause.
Both can be reasonable. A company unprepared for a crisis handles the crisis worse. But the scenarios Axios describes are, per the report, about the revolt rather than the attack. That’s a difference, and it says something about which kind of risk currently feels more pressing in those rooms.