The AI 'Slowdown' Is an Antitrust Mess

The so-called AI slowdown isn't a technical deceleration: it's an antitrust maze that the labs themselves helped build. By framing their pauses as a 'slowdown' rather than an industry-wide push for better security standards, companies like OpenAI, Google, and Anthropic walked straight into a regulatory problem that's just beginning.

How we got here

Since 2023, major labs have called for moratoriums and signed letters about existential AI risk. That talk sounded like a cartel: the same players dominating the market were asking to slow things down. Regulators took note, and the FTC started scrutinizing any coordination between competitors.

The irony is that safety was the argument, but the effect was protecting market share. Startups couldn't afford pauses; giants could. That turned a legitimate concern into a barrier to entry.

The details that complicate it

In January 2025, the FTC opened an inquiry into investment deals between Microsoft, Amazon, and Google with AI startups, looking at whether those alliances bypass antitrust review. Meanwhile, the European Commission questioned several labs about their 'responsible pause' practices. And in the US, the Department of Justice is already examining whether voluntary safety standards serve to coordinate pricing or compute access.

• OpenAI, Anthropic, and Google DeepMind control more than 70% of frontier models.
• Their self-imposed pauses coincided with API price hikes.
• Several governments now demand transparency on who decides what's 'safe'.

The regulatory message is clear: if you use safety as a shield, you have to prove it's not a cartel.

Analysis: who wins and who loses

Antitrust lawyers win; the labs lose room to maneuver. The 'slowdown' narrative gave them a public excuse to slow smaller competitors without bearing the cost of innovation. My take: the real slowdown isn't technical, it's legal. And regulators won't accept 'trust us' as an answer.

Implications for the industry

What's coming is more auditing, not less AI. We'll likely see mandatory safety standards instead of voluntary ones, and self-imposed pauses could become evidence in antitrust cases. Startups might benefit if rules level the playing field, but only if they survive compliance costs.

The 'slowdown' was a PR move; now it's a regulatory file. Did anyone at those labs think calling it a 'pause' would sound less suspicious than 'cartel'?

Source: Wired