Anthropic co-founder Jack Clark has warned that the rapid advancement of artificial intelligence is approaching a critical point where systems could begin improving and operating with minimal human involvement, urging governments and society to develop stronger controls to manage its development.
Speaking to BBC Newsnight, Clark said the AI industry is currently accelerating without sufficient safeguards to slow it down if necessary. “You want the option to be able to take your foot off the gas and put your foot on the brake,” he said. “Right now, it’s like the AI industry has a gas pedal, but it doesn’t have a brake pedal.” He argued that as AI systems grow more powerful and influential, public policy must evolve to ensure humans remain firmly in control.
Clark also called for clearer regulatory frameworks, comparing the current stage of AI development to earlier industrial eras that eventually required structured oversight. He said governments will need to build systems that give society confidence in AI’s safety and governance, rather than relying solely on the judgement of private companies leading the sector.
He revealed that Anthropic’s chatbot Claude is already generating around 80% of its own code, and suggested that figure could potentially reach full autonomy within the next two years. According to Clark, such a development would carry major consequences for the technology industry and beyond.
While acknowledging the benefits of AI, Clark also highlighted risks including economic disruption and job displacement as AI “agents” become capable of performing complex tasks independently. He suggested that recent workforce reductions across major technology companies have, in part, been linked to growing reliance on AI systems.
Clark said Anthropic’s intention in speaking publicly about these risks is not commercial positioning, but transparency about what developers are observing inside leading AI labs. He added, “I am worried for my kids if we as a society don’t have a serious conversation about what the implications of AI’s continued advances mean,” stressing both the opportunities and dangers of the technology.
Despite growing calls for tighter regulation, Clark noted that leading AI companies, including Anthropic, OpenAI and Google, have not agreed to pause development, and recent policy moves in the United States remain largely voluntary in terms of safety testing requirements.
He advised younger generations to adapt by developing broad, creative skills rather than focusing only on technical paths, suggesting that curiosity and interdisciplinary thinking may become increasingly valuable in an AI-driven economy.
