AI could slip beyond human control, Anthropic co-founder warns

ALBAWABA -- Anthropic co-founder Jack Clark has warned that artificial intelligence could advance beyond human control, urging governments to cooperate on safeguards before the opportunity narrows.

Speaking at the IMF and World Bank annual meetings in Bangkok on Saturday, October 17, Clark called for international coordination on "unusual and serious rules," according to Reuters.

His warning concerned the direction of AI development, rather than an announcement that existing systems had escaped human oversight.

The stakes extend to jobs and economic growth. Anthropic's published research explores several possible futures, ranging from gradual productivity gains to a profound transformation driven by increasingly autonomous systems.

The company stresses that these are conditional scenarios, rather than firm predictions. Outcomes depend on how quickly AI improves and how widely businesses adopt it.

In its most extreme scenario, AI becomes more productive than people at most knowledge-work tasks, dramatically increasing economic output while reducing employment opportunities in those occupations.

More moderate scenarios show smaller gains, with unemployment remaining within historical ranges.

Anthropic's research highlights a difficult possibility: a wealthier economy could still leave many workers worse off unless the benefits are widely shared. That makes the debate about controlling AI inseparable from decisions about who gains from its development.