The co-founder of Anthropic, the artificial intelligence safety company, has issued a stark warning: AI development must not proceed without robust human oversight, and the United Kingdom is uniquely positioned to lead on regulation. Speaking at a tech policy conference in London, the executive argued that the rapid pace of AI advancement poses existential risks if left unchecked, calling for a “human-centred” approach to governance.
The remarks come amid a global race to regulate AI, with the European Union’s AI Act nearing finalisation and the US Congress still debating frameworks. The co-founder stressed that Britain, with its established legal traditions and tech sector, should seize the opportunity to set a global standard. “We need a regulatory environment that is adaptive but firm, one that puts human welfare at the core of algorithmic design,” they said.
Anthropic’s warning is particularly urgent given the rise of large language models and generative AI. The co-founder pointed to recent incidents where AI systems have exhibited biased or unpredictable behaviour, underscoring the need for continuous human oversight. “We cannot automate away accountability,” they said. “Every line of code that impacts people’s lives must be scrutinised by humans who understand its societal implications.”
The call for British regulation is rooted in the country’s potential to balance innovation with caution. The co-founder highlighted the UK’s Financial Conduct Authority as a model for AI regulation: agile, principle-based, and focused on outcomes. They urged the government to create a dedicated AI watchdog with the power to audit algorithms, enforce transparency, and halt deployments that pose public harm.
However, critics argue that heavy-handed regulation could stifle innovation and drive AI companies to less restrictive jurisdictions. The co-founder countered that regulation, done right, can foster trust and investment. “Users need to know that the AI they interact with is safe, fair, and accountable. That trust is the bedrock of sustainable growth,” they said.
The debate also touches on digital sovereignty. As AI systems become embedded in critical infrastructure from healthcare to finance, the question of who controls and oversees these systems becomes existential. The co-founder warned against outsourcing that oversight to a few powerful corporations. “We need democratic governance of AI, not corporate governance,” they asserted.
The urgency of the warning is amplified by the accelerating capabilities of AI. Just last week, a new model from a leading lab demonstrated emergent reasoning skills that surprised even its creators. For the Anthropic co-founder, this is a clear signal that the window for meaningful regulation is closing. “We are in a race against time to build guardrails before these systems become too complex to control,” they said.
The UK government has yet to formally respond, but insiders suggest the Department for Science, Innovation and Technology is drafting a white paper on AI governance. The co-founder’s intervention adds pressure to move from consultation to legislation. With the next election looming, the clock is ticking.
In the meantime, Anthropic is implementing its own safeguards: human-in-the-loop reviews for all high-risk applications, open-source transparency reports, and a public commitment to ethical AI. But the co-founder insists that voluntary measures are not enough. “We need legally binding rules, not just goodwill,” they said. “The consequences of inaction are too grave to leave to chance.”
