Jack Clark, co-founder of Anthropic, is worried about how fast AI is growing. Jack feels we don’t have enough control over it right now and believes we need stronger rules—and even a way to slow things down if needed.
Speaking to BBC Newsnight, Clark used a simple but powerful comparison: imagine a car that can speed up but has no brakes. That’s how he sees the current state of AI development.
“You want the option to take your foot off the gas and hit the brake,” he said. “Right now, AI has the gas pedal, but no brake pedal.”
Governments and policymakers need to take necessary steps before it becomes a disaster. As AI becomes more advanced, it’s critical to ensure that humans remain in control and that people can trust these systems.
One of his biggest concerns is how fast AI is improving its own capabilities—especially in software development. Jack unleashed Anthropic's achievement in Claude, claiming that Claude itself is already running on code that is largely written by AI itself. Around 80% of its code is AI-generated today, and Clark believes In near future this number could rise to 100%.
Jack also highlighted the impact on jobs cut due to AI. As Artificial Intelligence becomes more advanced, they may start handling tasks with zero human involvement, potentially replacing certain roles over time.
Beyond that, Anthropic has also warned about a future possibility called “recursive self-improvement.” This is where AI systems could eventually become capable of improving themselves—designing and developing their own, more advanced versions.
The company stresses that we’re not there yet, and it’s not guaranteed to happen. But if progress continues at the current pace, it could arrive sooner than many institutions are prepared for.
