US Government Cannot Verify AI Laboratory Claims, According to Security Experts

According to a new survey published by the Institute for Security and Technology (IST), national security experts state that the U.S. government lacks the capacity to verify the claims of leading AI laboratories.

Scope of the Research

The survey report published by the Institute for Security and Technology (IST) reveals the views of national security experts on artificial intelligence risks. Conducted between late April and mid-July, the study involved 111 experts from various national security fields.

Lack of Verification Capacity

According to the report, the government lacks the ability to verify claims made by leading artificial intelligence laboratories regarding their products. Chronic underinvestment in testing and evaluations, along with a lack of independent processes, are cited among the primary reasons for this situation.

Cybersecurity and Risks

While experts anticipate that artificial intelligence systems will improve cyber defense, they also note that in the short term, they are more helpful to attackers. The ability to discover and exploit vulnerabilities is ranked as the greatest cybersecurity risk.

Regulatory Approaches

Experts hold differing views on the regulation of artificial intelligence. Approximately twenty-five percent support cross-sector federal requirements, thirty-one percent support sector-specific requirements, and thirty-six percent advocate for a hybrid approach.