Anthropic's Chief Executive Says He Underestimated How Fast Artificial Intelligence Would Reach the Global Economy

The chief executive of the artificial-intelligence company Anthropic said he did not appreciate the speed with which the technology would come to underpin the global economy, in remarks reported by Al Jazeera. Coming from an executive who has consistently argued for caution, the statement is an acknowledgement that the industry's own forecasts have repeatedly understated the rate of adoption.

The Financial Times published a column on the same day arguing that reducing the speed of innovation will be hard both practically and politically, but that it is the right decision. Its central point is that the standard regulatory posture, which is to wait for demonstrated harm and then legislate, does not work when the harm arrives faster than the legislative cycle.

Governments deploying the technology are applying a different standard. Indonesia's deputy minister of communications and digital affairs, Nezar Patria, said the government measures the success of artificial intelligence in public services by real impact rather than by the number of systems deployed. That is a procurement standard rather than a safety standard, but it addresses the same gap between announced capability and delivered value.

For investors the relevant question is where the constraints on the sector will come from. So far they have come from inside the industry, through published commitments, evaluation frameworks and capability thresholds that the leading laboratories negotiate among themselves. Regulators in most jurisdictions are still drafting rules. A sector whose operating rules are written by its own largest participants carries a specific risk: the rules can tighten quickly and without a legislative process, and a company that ships a capability its peers have agreed not to ship pays a reputational cost that shows up in enterprise contracts.