loading
loading
Liz’s view Picture three competing CEOs, worried that their product might kill people, hopping on a conference call to discuss shutting it all down for a while. Then the lawyers step in. That was March 2020, when the CEOs of General Motors, Ford, and Fiat Chrysler tried to coordinate production halts over Covid-19 fears, only to be warned by their lawyers that doing so could violate antitrust rules. The irony was obvious: “Laws meant to protect workers — by preventing companies from colluding to decrease wages or increase hours — could now put them in harm’s way,” I wrote in my book about the pandemic’s economic toll. (Proof that I was liberally using em dashes before the arrival of AI; do not Pangram me.) Legal concerns were swiftly dismissed, Detroit’s factories went dark, and people were kept out of harm’s way. Today the danger is swarms of malevolent AI agents, the CEOs in question run frontier labs, and legal niggling risks holding up an urgent safety conversation. Dario Amodei’s request for a “narrow” antitrust waiver to allow a coordinated slowdown of AI model development set off a debate about whether the industry is weaponizing doom to pull up the drawbridge behind it. That’s silly. These guys have been talking about AI safety for a long time. Nobody woke up last week, looked at their IPO plans, and decided existential risk was good marketing. But an “AI OPEC” isn’t really in the cards, anyway. Mark Zuckerberg certainly wants no part of one, and cutthroat competition seems hard-coded into tech executives, who are rich enough not to need to win but wake up every morning trying to destroy each other anyway. Whatever cartel-like behavior does emerge will probably look less like coordinated price-fixing and more like the airline industry, which competes on reliability and bundled extras as the underlying product commoditizes. (Notably, airlines never compete on safety; crashes are bad for everyone.) Even the real OPEC isn’t especially effective these days. It couldn’t stop the UAE from walking out, and it doesn’t include three of the world’s four largest oil producers. More to the point, AI looks increasingly like a pro-competitive force — lowering the cost of starting almost any kind of business. The lawyers should probably stay out of this one. Notable - Microsoft CEO Mustafa Suleyman criticized Anthropic’s approach to training Claude on ideas related to consciousness, Reuters reported.
document fetched 2026-09-18 02:55 · canonical https://semafor.com/article/09/17/2026/the-idea-of-an-ai-opec-will-likely-not-materialize