AI Requires a 'Brake Pedal,' Says Anthropic Co-Founder

Jack Clark, co-founder of Anthropic, has issued a stark warning regarding the trajectory of artificial intelligence, urging the development of mechanisms to halt or slow the technology’s rapid evolution. Clark cautioned that AI is approaching a threshold where it may begin developing autonomously, without direct human intervention.

Speaking to BBC Newsnight, Clark emphasized the necessity of having control options. "You want the option to be able to take your foot off the gas and put your foot on the brake," he explained. "Right now, it's like the AI industry has a gas pedal, but it doesn't have a brake pedal." He argued that society, guided by government policy, must retain authority over these increasingly powerful systems, which will inevitably exert a broader influence on the world. "The world needs to do some thinking and we need to eventually develop some new regulations that allow us to be confident in these systems," he stated.

The urgency of this concern is underscored by recent developments at Anthropic. Its widely used chatbot, Claude, currently operates on code that is 80% self-generated. Clark noted that reaching a state of 100% self-writing code is feasible within two years, a milestone he described as having "huge implications."

While Clark did not specify exactly how an AI "brake pedal" would be implemented, he drew historical parallels to the oil industry at the turn of the last century. He suggested that, much like the response to the oil boom, society will eventually establish a sensible regulatory framework. This approach, he argued, would provide public confidence in the technology’s benefits while insulating the system from the whims of individual corporate leaders. "That's clearly where we end up here," Clark concluded.

This call for caution contrasts with Anthropic’s recent reception of US President Donald Trump’s executive order on AI, which the company welcomed for its relatively hands-off approach. The directive did not mandate government safety testing for AI firms, leaving such efforts voluntary. Furthermore, major players in the sector, including Anthropic, OpenAI, and Google, have not committed to pausing their own research and development efforts.

Despite these warnings, Anthropic continues its aggressive growth. Founded five years ago by CEO Dario Amodei, Clark, and other executives, the company is preparing for an initial public offering. It is expected to become one of the first major listings by a newer AI firm and potentially one of the most valuable in history, with private investors estimating its valuation at nearly $1 trillion (£745bn).

Clark clarified that his motivation for highlighting the rapid advancements within AI labs is not to enhance Anthropic’s reputation among clients. Instead, he aims to "tell the world what we're seeing inside these companies with this unusual technology." Since its inception, Anthropic has positioned itself as a vocal critic of AI risks. Notably, the company entered into a public dispute with the US Department of Defense, citing concerns that its tools could facilitate mass surveillance of Americans and autonomous warfare.

Clark expressed deep personal concern about the future, telling Newsnight, "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." He acknowledged that while AI offers significant benefits, it also carries substantial risks, particularly regarding economic disruption. He pointed to the rise of "agents"—autonomous AI bots capable of performing routine tasks—as a threat to certain job sectors. This fear is reflected in recent mass layoffs across major tech firms, which have cited AI’s ability to replace the work of hundreds or even thousands of software engineers.

However, Clark suggested that human creativity may remain a competitive advantage. He questioned whether AI can truly be creative, noting a lack of evidence to support that claim. "At Anthropic, we're now limited more by the ability to generate good ideas than the ability to do the engineering to turn those ideas into reality," he said, implying that individuals with strong, innovative ideas may still hold an edge in an AI-driven economy.

Source: BBC News Generated at: 2026-06-04 21:53:25 UTC