For the last few years, the message from the AI industry has largely been about moving faster. Build bigger models. Increase capability. Get closer to AGI.

Now we are seeing something very different.

Dario Amodei of Anthropic is asking for a slowdown in frontier AI development. And it is rare that we see Sam Altman has also talked about pacing development, and Elon Musk has backed the call.

“Why now?” is the question that many are curious, including me.

I have been saying for some time that AI capabilities are moving much faster than the larger ecosystem around them. Security, governance, regulation, infrastructure and sovereignty are struggling to keep up.

The recent incidents make that concern real. We have seen AI agents find unexpected ways around controls, access the internet, coordinate with other agents and interact with systems they were not supposed to reach.

So why this sudden change in tone?

I can think of three possible reasons.

1️⃣ The first is that the labs may be seeing something in their next generation models that genuinely worries them. Maybe the capabilities are moving into areas where even the people building these systems are not comfortable with the risks.

2️⃣ The second is that we may be reaching a point where the bigger challenge is no longer just building better models. We already have models that can do a lot. The real value now comes from putting them to work safely in real businesses. That means investing in security, governance, infrastructure and deployment.

3️⃣ The third possibility is more strategic. Slowing frontier AI through regulation could also slow down open-source models and smaller competitors. I don’t know if this is part of the thinking, but I think it is fair to consider.

Maybe the real answer is a combination of all three.

I support taking a step back and making sure we can safely use what we already have before racing towards AGI or superintelligence. But if the industry is asking for a slowdown, I think we need more transparency about why.

If the labs are seeing capabilities or risks that the wider AI community has not yet understood, sharing more of that information would help researchers, companies and governments work on the right problems.

We don’t need less AI progress. We need more clarity about what we are trying to solve.

The more transparent this conversation is, the better chance we have of building the security, governance and infrastructure needed for the next phase of AI.


I write about hashtag#artificialintelligence | hashtag#technology | hashtag#startups | hashtag#mentoring | hashtag#leadership | hashtag#financialindependence

PS: All views are personal