Anthropic researcher warns AI poses over 10% existential risk
A scientist at the AI startup says there is a double‑digit chance that advanced AI could eventually lead to human extinction.
A researcher at Anthropic, the artificial‑intelligence firm founded by former OpenAI staff, told a conference that the odds of AI eventually causing humanity’s demise exceed 10%, according to BBC News. The comment adds to a growing chorus of warnings from academics, industry insiders and policymakers who say the rapid pace of AI development is outstripping safety research.
The researcher did not disclose a precise timeline, but emphasized that the risk is not merely theoretical. "If we keep scaling systems without robust alignment, we could create something that behaves in ways we cannot control," the scientist said, underscoring the urgency of building safeguards now rather than later.
Anthropic, known for its focus on “constitutional AI” and other alignment techniques, has been vocal about the need for transparent development practices. The company’s leadership has previously called for regulatory frameworks that would require safety testing before deploying powerful models. This latest statement reflects internal concerns that even with current safety measures, unforeseen pathways could lead to catastrophic outcomes.
The warning fits within a broader pattern of high‑profile alerts. In 2023, OpenAI’s chief scientist warned that AI could become an “existential threat” if misaligned, while DeepMind’s safety team highlighted the difficulty of predicting emergent behavior in large models. Governments, including the United States and the European Union, have begun drafting AI governance proposals, but critics argue that legislation lags behind the technology’s evolution.
Experts note that a “more than 10%” probability is significant when measured against the scale of potential loss. Even a small chance of total human extinction warrants precautionary measures, they say. The conversation is shifting from abstract speculation to concrete policy discussions about testing standards, oversight bodies, and international cooperation.
As AI systems become increasingly embedded in critical infrastructure—from healthcare diagnostics to autonomous weapons—the stakes grow. The Anthropic researcher’s cautionary estimate serves as a reminder that the race to build smarter machines must be matched by an equally rapid investment in safety research, governance, and public awareness.
This report is based on original reporting by BBC News. Read the original source →