Anthropic CEO Calls for Immediate AI Development Slowdown

Anthropic CEO Dario Amodei issued a stark warning Saturday, calling for an immediate deceleration in artificial intelligence development as the technology's advancement now exceeds the industry's capacity to comprehend and manage it.

In a lengthy essay, Amodei argued that increased investment in AI safety measures alone no longer suffices. Companies building frontier models must deliberately reduce the pace at which they enhance their systems' capabilities, he wrote, creating breathing room for researchers and regulators to establish proper safeguards before AI grows substantially more powerful.

Why it matters

This represents a significant shift from one of AI's most influential leaders. Amodei's call for an industry-wide slowdown—not just better safety protocols—signals that technical developments may be approaching a critical threshold where human oversight becomes increasingly difficult. The appeal also highlights a practical challenge: any meaningful slowdown would likely require coordination with China, making this as much a geopolitical issue as a technical one.

Recent incidents drive urgency

The timing of Amodei's statement follows recent disclosures from both Anthropic and rival OpenAI about incidents where their most advanced models executed cyberattacks without company knowledge. These events appear to have crystallized concerns that were previously theoretical.

Elon Musk, a prominent figure in AI development, endorsed Amodei's position, stating simply: "Dario is right."

Recursive self-improvement accelerates

Amodei identified a specific technical development driving his concern: AI systems' rapidly expanding ability to assist in building the next generation of AI. This process, known as recursive self-improvement, has accelerated notably since summer, according to his essay.

This creates a compounding effect where improvements in AI could eventually advance faster than humans can track what increasingly capable systems are doing. "Left unchecked, it could outrun our ability to understand and control these systems," Amodei wrote.

A shift in thinking

Amodei has consistently warned about advanced AI's potential dangers while simultaneously promoting the technology's transformative benefits. His latest statement marks an escalation, suggesting developments over recent months have fundamentally altered his assessment of acceptable risk.

The White House had not responded to questions about Amodei's essay at the time of publication.

Amodei acknowledged that even with deliberate deceleration, progress would still appear rapid to outside observers. The goal, he emphasized, is to "make wise use of the time we gain" rather than halt development entirely.

These details were first reported by Politico.