Jack Clark, co-founder of AI company Anthropic, has stressed the urgent need to introduce mechanisms that can slow down artificial intelligence development, warning that the technology is approaching a stage where it might evolve independently from human guidance. He compared the current AI industry to a vehicle with a gas pedal but no brake, emphasizing the necessity to have the option to "put your foot on the brake" to avoid uncontrolled advancement.
Clark highlighted that government intervention must play a central role in maintaining human oversight as AI systems grow more powerful and begin to affect society on a wider scale. He called for thoughtful reflection and the establishment of new regulations to build confidence in the safety and reliability of these technologies. Anthropic’s own chatbot, Claude, already runs largely on self-generated code—Clark said about 80% of it—which could realistically reach full autonomy within a couple of years, an evolution with far-reaching consequences.
While Clark stopped short of detailing how this “brake pedal” could be engineered, he drew a historical parallel with how society responded to the early 20th-century oil boom. Governments implemented sensible policies and regulatory frameworks to harness oil’s benefits safely, removing concerns about the individuals controlling the industry. Clark suggested the AI sector requires a similar approach to ensure societal trust and control.
This call for regulated AI growth comes even as US President Donald Trump recently signed an executive order on artificial intelligence, which Anthropic welcomed but which did not impose mandatory safety testing on companies, leaving such oversight voluntary. Major AI companies, including Anthropic, OpenAI, and Google, have not committed to pausing their development efforts. Meanwhile, Anthropic is preparing for a major public listing that could make it one of the most valuable new entrants on the stock market, with private valuations nearing $1 trillion.
Clark underlined that Anthropic’s transparency about AI’s escalating capabilities is driven by a desire to inform the public rather than to boost company image. Since its founding by several former OpenAI employees in 2021, Anthropic has positioned itself as vocal about AI risks, including publicly challenging the US Department of Defense over concerns related to surveillance technologies and autonomous weapons.
The urgency in Clark’s warnings reflects a broader debate over how society should balance innovation with precaution, especially given AI’s increasing potential to operate with reduced human intervention. His comments invite policymakers, industry leaders, and the public to engage in a serious discourse about the future of AI governance.