Automattic’s 33-Hour Coup, and can AI labs police themselves?

TechCrunch’s Equity episode frames two power stories at once. On AI: a week after an Anthropic researcher’s doomsday warning, Dario Amodei’s plan to pace the frontier via independent evaluators and coordination among labs in democratic countries has drawn support and pushback from Nvidia’s Jensen Huang. Hosts Kirsten Korosec, Anthony Ha, and Sean O’Kane ask whether companies can agree on what slowing down means and who polices it.

On infrastructure: WordPress parent Automattic’s boardroom coup—Matt Mullenweg’s 33-hour ouster, golden parachute deals, and the risk that drama gives him even more control over technology that powers much of the web. The show also covers May Mobility’s SPAC path to more than $300 million in potential funding and DoorDash’s $425 million investment in Wonder amid the race to automate food from order to delivery.

Self-regulation talk lands next to a reminder that control of platforms still turns on boards, parachutes, and capital—not essays alone.

We’re listening to the AI pacing debate with one ear on Automattic—because governance failures look the same whether the asset is a model or the CMS under half the internet.

Amodei’s evaluator pitch needs referees with escalation rights. Huang’s pushback shows voluntary clubs fracture. Equity is right to refuse “slowdown” as a vibe.

The Automattic thread is the parallel lesson: 33 hours of board chaos around WordPress’s parent is a stress test for how fast concentrated control can reshuffle. If AI labs want to claim they can police themselves, they should explain what happens when a CEO, a board, or a chip partner defects mid-pact.

Useful AI still deserves room to ship. Self-policing without consequence is theater. I’m watching named evaluators, written defection costs, and whether Mullenweg’s aftermath concentrates more web infrastructure power—not less.

TechCrunch published the Equity podcast page on September 18, 2026, with Korosec, Ha, O’Kane, and Loconsolo.