Anthropic's Jack Clark Warns AI Could Write 100% of Code Within 2 Years
Updated
Updated · BBC.com · Jun 4
Anthropic's Jack Clark Warns AI Could Write 100% of Code Within 2 Years
1 articles · Updated · BBC.com · Jun 4
Summary
80% of Claude’s code is already AI-written, and Jack Clark said reaching 100% within two years is plausible, arguing the industry needs a regulatory “brake pedal” before systems advance beyond human control.
Clark said governments must create rules that let people keep control of increasingly powerful AI, warning that autonomous agents could disrupt jobs and the broader economy.
Anthropic has not proposed a concrete mechanism to slow research, and the company this week welcomed Donald Trump’s relatively hands-off AI executive order, which kept government safety testing voluntary.
Nearly $1 trillion is Anthropic’s estimated private valuation as it prepares for a potential IPO, underscoring the tension between rapid commercial expansion and its long-running public warnings about AI risk.
As AI writes its own code, can any regulation truly prevent its misuse in autonomous warfare and mass surveillance by world powers?
Anthropic urges caution while pursuing a $1T valuation. Is this a genuine safety plea or a strategic move to dominate the market?
If AI automates the jobs that teach human judgment, how does the next generation learn to lead in an AI-driven world?
AI Surpasses Human Coders: Anthropic’s Claude Mythos and the Urgent Cybersecurity, Workforce, and Governance Challenges of 2026
Overview
The report highlights the rapid evolution of AI, focusing on Anthropic's Claude Mythos Preview, which marks a turning point in AI's coding abilities and cybersecurity impact. This advanced, unreleased model surpasses most human experts in finding and exploiting software vulnerabilities, signaling a new era where AI acts as an active agent in digital security. Its powerful capabilities have sparked serious reflection within the AI community about both the opportunities and risks, especially as AI begins to reshape offensive and defensive cyber strategies. Anthropic’s cautious approach to releasing this technology underscores the urgent need for robust governance and responsible deployment.