
"On December 11 th, the Trump Administration released a new executive order (EO) that has completely changed the long-term trajectory of AI regulation in the United States, effectively taking AI regulatory power away from the states and putting it in the hands of the federal government. In the short-term, this move has created disruption and risks that organizations will have to insulate themselves against."
"Before the EO on December 11th, various states had passed their own AI laws, leading to an uneven and sometimes contradictory web of regulations across the country. California, for example, passed a series of laws in recent years that required AI companies to audit the safety of their models, limit discrimination, and disclose the presence or use of AI in certain settings."
"Below, I've outlined what this development means for organizations that use AI and the tactics that leaders can use to prepare for these changes, both over the short- and long-term. In the same way that international and harmonized standards allow organizations to understand the expectations of regulators, a federal requirement, rather than potentially fifty competing and conflicting standards, will ultimately be more predictable and efficient for compliance."
A December 11 executive order transfers AI regulatory authority from states to the federal government, shifting the long-term trajectory of AI regulation. In the short term, the order creates disruption and increased legal and compliance risks that organizations must insulate against. The order aims to nullify state laws as a first step toward centralized, nationwide AI legislation, which should reduce conflicting standards and improve safety and efficiency. State approaches vary: California mandates model safety audits, discrimination limits, and AI disclosure; Texas adopted different, less stringent rules; South Dakota, Colorado, and Utah have also enacted laws. A single federal standard would improve predictability for compliance.
Read at Securitymagazine
Unable to calculate read time
Collection
[
|
...
]