Anthropic and OpenAI Just Agreed to Slow Down. Your AI Adoption Plan Was Never About Their Pace.

This week, the people who build the most advanced AI models in the world asked each other to slow down. If you lead a team that is two months into rolling out AI, or still deciding whether to start, that headline probably landed with a jolt: are we moving too fast too.

On September 12, Anthropic CEO Dario Amodei published an essay titled "We Must Pace the Frontier." His argument: AI capability is advancing faster than the industry's ability to understand and control what it has built, and labs should deliberately slow the rate at which they push new capability out the door. He proposed three concrete steps: independent third party evaluations with real access to model internals, coordination across labs on safety thresholds, and cooperation at a global level.

Hours later, OpenAI CEO Sam Altman agreed publicly. He said pacing the frontier had already been a live discussion inside OpenAI and committed to the same kind of independent evaluator access Amodei proposed. Elon Musk added a one line endorsement on X: "Dario is right." Three competitors, usually careful not to hand each other a talking point, lined up behind the same message within a single afternoon.

The easy read is that this is a warning light for anyone using AI at work: if the people building it are worried, maybe we should be worried too, maybe we should pause the rollout, wait for calmer, safer models.

That reading mixes up two different clocks. Amodei's essay is about the pace of frontier training, how fast the next generation of models gets built and shipped by a handful of labs with access to enormous compute. It has almost nothing to do with the pace at which your finance team learns to use the tools that already exist, or whether your customer service process has a person checking AI drafted replies before they go out. Those are your clocks, not theirs. Nothing in this week's news changes what your organisation should be doing with the AI capability that has already been sitting on the table for the past year.

Separate the two conversations explicitly. When capability headlines land, ask whether this changes what you should build, or whether it just makes people nervous. Most weeks, the honest answer is the second one.

Treat the call for external evaluation as a cue for internal evaluation, not a substitute for it. If the labs building frontier models are asking for outside checks on what they ship, that is a strong argument for your own team having a plain, written answer to who reviews this before it goes to a client, for whatever AI touches your work today.

Use the moment to open the governance conversation without alarm. You do not need a compliance department to have one page that says which tools are approved, who owns the account, and what a human always checks. This is the kind of clarity the Alignment part of AI adoption depends on, and most SMEs still do not have it written down.

Do not wait for a safer model to start. The gap that matters for most companies right now is not a capability ceiling, current tools already do far more than most teams use, it is clarity about what problem you are actually solving. That is a Meaning question, not a technology one.

If every AI lab froze development today, exactly where it stands this week, how much would actually change inside your organisation. For most leaders, the honest answer is: not much, because the constraint was never the technology. Amodei and Altman are having a conversation about their industry. You are still free to have yours.

Try this week

Pick one place in your business where AI output currently reaches a customer or a decision with no named person checking it first: an email draft, a report summary, a proposal. Write one sentence naming who reviews it before it ships. That sentence is the start of the governance your organisation needs regardless of what the frontier labs do next.

Sources

• Sam Altman backs Anthropic CEO's call to slow down the global AI race, NBC News
• Anthropic, OpenAI CEOs call for slowdown in AI development, Axios