President Joe Biden has signed an executive order outlining regulations for generative AI. This directive comes as a precursor to any formal legislation from lawmakers, emphasizing the urgency of setting standards in the rapidly evolving AI landscape.
Key objectives of the order
The executive order sets eight primary goals, including the establishment of AI safety and security standards, privacy protection, and the promotion of equity and civil rights — as well as advocating for consumers, students, and patients, supporting the workforce, fostering innovation, and ensuring the government’s responsible use of AI.
According to reporting by The Verge, several government bodies, including the National Institute of Standards and Safety, have been entrusted with specific roles. NIST, for instance, will develop standards for “red teaming” AI models before their public release. Meanwhile, the Department of Energy and Department of Homeland Security are to address AI’s potential threats to infrastructure and assess its cybersecurity risks.
Implications for AI developers
The order mandates that prominent AI developers, such as OpenAI’s ChatGPT and Meta’s Llama, disclose their safety test outcomes. The Biden administration has clarified that while they won’t recall existing AI models, these models must still adhere to existing anti-discrimination regulations.
To further safeguard user privacy, Biden urged Congress to enact privacy laws and has expressed commitment to developing “privacy-preserving” techniques.
Addressing discrimination and job displacement
The order takes a firm stance against AI-induced discrimination. It seeks to ensure fairness in AI applications, especially in sensitive areas like sentencing and surveillance. Additionally, guidelines will be provided to federal benefits programs, landlords, and contracts to prevent AI from intensifying discrimination.
Recognizing AI’s potential impact on the job market, the order also directs agencies to study AI’s influence on employment. The administration is keen on nurturing talent in the AI domain, prompting the launch of a National AI Research Resource for students and researchers.
The Biden administration introduced an AI Bill of Rights, later evolving into agreements with tech firms Google, Nvidia, and Adobe.
However, it’s crucial to note that an executive order isn’t permanent and typically lasts only through the president’s tenure. While lawmakers are still deliberating on AI regulations, some anticipate AI-related laws to be enacted by year-end.
Credo AI founder Navrina Singh views the executive order as a positive step, signaling the U.S.’s commitment to generative AI.