Anthropic Calls for AI Development Pause as IPO Looms

Anthropic has issued a stark warning about the pace of AI advancement while simultaneously preparing for what could be one of the largest technology IPOs in history, raising questions about the timing and intent behind its call for an industry-wide development pause.

In a blog post published Thursday, Anthropic researchers Marina Favaro and Jack Clark revealed that more than 80% of code now merged into the company's codebase is written by Claude, its AI assistant. Engineers are shipping roughly eight times as much code per quarter compared to pre-2025 levels, according to the post first reported by Fortune.

The authors warn this trajectory points toward "recursive self-improvement"—a threshold where AI systems autonomously design, build, and train their own successors without meaningful human oversight at each step. While Anthropic says this capability hasn't arrived yet, the company believes it "could come sooner than most institutions are prepared for."

Why it matters

The call for a coordinated pause comes at a pivotal moment for both Anthropic and the AI industry. If major labs agreed to halt development simultaneously, it could reshape competitive dynamics and regulatory approaches. But the proposal's arrival just days after Anthropic filed confidential IPO paperwork—and as it reaches a $965 billion valuation—has prompted skepticism about whether safety concerns or market positioning are driving the message.

The pause proposal's conditions

Anthropic isn't advocating for a unilateral halt. The company says a pause should only occur if "multiple well-resourced labs at or near the frontier, in multiple countries, agree to stop under the same conditions." That would presumably include OpenAI, Google, xAI, and Meta—all of which are racing to develop more powerful models, with several also preparing for public offerings.

The likelihood of such coordination appears slim given the competitive stakes and looming IPOs across the sector.

A reversal on safety commitments

The timing has drawn scrutiny partly because Anthropic recently walked back a key safety pledge. In February, the company overhauled its Responsible Scaling Policy, removing its central commitment to never train an AI system unless it could guarantee adequate safety measures in advance. Chief science officer Jared Kaplan said at the time that stopping unilaterally "wouldn't actually help anyone" while competitors continued development.

Critics, including Trump adviser David Sacks, have accused Anthropic of using safety rhetoric as "regulatory capture"—an attempt to slow rivals under the guise of responsible development. The company rejects this characterization.

What comes next

Anthropic says it plans to convene policymakers, researchers, and civil society groups in coming months to address questions around recursive self-improvement. The company emphasized that "people outside AI companies should be involved in this deliberation."

Whether the proposal represents genuine safety advocacy or pre-IPO positioning, Anthropic's warning underscores a broader tension: the window for establishing guardrails on increasingly autonomous AI systems may be closing even as competitive and financial pressures accelerate development.

These details were first reported by Fortune.