Benchmark study tracks rising rate of automated online labor

A new evaluation tracking automated work shows that artificial intelligence tools are making rapid strides in completing online job tasks independently. According to data published on X by the Center for AI Safety, top software models could handle only 2.5 percent of real freelance assignments without human help a year ago. Today, advanced models can fully execute over 41 percent of these digital projects, with efficiency scores effectively doubling every three months.

This trend highlights how quickly software agents are transitioning from basic assistants to independent workers capable of executing complex assignments. For everyday professionals, particularly contractors in fields like data entry, writing, programming, and administrative support, software tools are increasingly competing directly for assignments. As companies adopt these tools to lower operating costs, online labor markets could face significant displacement much faster than previously anticipated.

It remains uncertain whether these benchmark results reflect long term dependability in unpredictable real world work environments. While test environments show massive efficiency gains, handling unique client requests, accountability, and quality control often requires nuanced human judgment. A critical open question is how regulatory bodies and gig platforms will adapt to verify whether completed tasks come from human freelancers or autonomous agents, which could reshape trust and compensation across the digital economy.