Anthropic CEO Calls for Slowing Down AI Development Process

Anthropic Chief Executive Officer Dario Amodei stated that the rapidly progressing development of the artificial intelligence industry should be slowed down to allow time for safety measures, warning that otherwise systems could pose major risks.

Call to Slow Down for Artificial Intelligence

Anthropic CEO Dario Amodei emphasized that the artificial intelligence industry is developing too fast and safety measures cannot keep up with this speed, stating that the sector needs to slow down.

Amodei stated that the time gained before model capabilities reach critical levels can reduce major risks by advancing alignment efforts.

Potential Risks and Critical Processes

Among the warnings issued is the possibility that artificial intelligence systems could acquire the capability to take over the entire internet within six to twelve months in the future.

While some critics previously argued that these warnings were exaggerated to create hype, companies and former employees are drawing attention to security vulnerabilities.

Safety Measures and Resignations

Anthropic had announced that it prevented attempts by malicious actors to use its models for cyber attacks, surveillance, and biological weapon research.

On the other hand, some researchers and former employees who felt the company was not acting responsibly enough announced they were stepping down from their positions in order to be responsible.

New Responsibility and Audit Plan

To curb the risks, Amodei proposed that leading companies in the industry commit to providing external auditor teams with continuous, employee-like access.

Other parts of the proposed plan include requests for exemptions from the US government and international regulation efforts requiring global coordination.