Anthropic’s CEO just told the world to slow down.
This weekend, Dario Amodei published an essay calling for AI companies to deliberately pace how fast they push model capabilities forward — arguing that safety, oversight, and alignment work need to catch up before the industry pushes further. OpenAI’s Sam Altman and even Elon Musk publicly agreed within hours.
When the people building the most advanced AI in the world are asking for guardrails, it’s worth asking a smaller but very real question for the rest of us:
Are we moving at the right speed with AI in our own businesses?
At wonderIT.io, we see this trade-off on nearly every project. A client wants to ship an AI feature fast. We can absolutely move fast — but we’ve learned (sometimes the hard way) that skipping testing, code review, and proper QA to hit a deadline almost always costs more time later than it saves now.
One recent example: a client asked us to fast-track an AI-powered feature for a launch deadline. We pushed back on cutting our review process, added an extra QA pass instead, and shipped two days later than the original ask. The two-day delay was a footnote. The bugs we caught before launch weren’t.
Speed and safety aren’t opposites — but they do require intention. That’s true whether you’re training a frontier model or shipping a feature to production next Friday.
Curious where others land on this:
Do you think the software industry (not just AI labs) moves too fast, too slow, or about right when it comes to shipping with confidence vs. shipping fast?