The most ambitious effort by the government to establish regulations for this technology was initiated by President Joe Biden through an executive order on artificial intelligence set to be released next month. The primary objective of this order is to curb Silicon Valley’s tendency to launch products prematurely, emphasizing the importance of implementing best practices and standards for artificial intelligence designs.
Despite the comprehensive nature of the order, spanning 111 pages and addressing various topics such as industry standards and civic rights, there are two significant omissions that could undermine its effectiveness.
Firstly, the order fails to address the loophole created by Section 230 of the Communications Decency Act. This loophole poses a significant risk in terms of the proliferation of deep fakes, which includes manipulated video, audio, and image content leading to misinformation. While the order includes provisions for tagging and hashing AI-generated content to track its origins, the absence of such labels raises concerns about accountability.
Furthermore, the responsibility for hosting a substantial amount of AI-generated content lies with social media platforms like Instagram and X (formerly Twitter). The surge in deceptive content, particularly fake nude images, underscores the potential harm caused by such technology. However, under Section 230, platforms are largely shielded from liability for third-party content posted on their sites. This raises questions about the platforms’ incentive to remove AI-generated content, regardless of whether it is watermarked or not.
Addressing the issue of liability solely on the creators of AI content rather than the platforms distributing it may prove insufficient. Identifying content creators can be challenging, and they may operate outside legal jurisdictions or lack the financial means to bear responsibility. Despite the protection afforded by Section 230, platforms can still disseminate harmful content and even profit from it, particularly in the case of advertisements.
A bipartisan effort led by Sens. Richard Blumenthal (D-Conn.) and Josh Hawley (R-Mo.) aims to address this liability gap by proposing legislation that removes Section 230 immunity for claims related to synthetic intelligence. However, the question of apportioning blame between AI companies and hosting platforms remains unresolved in the proposed legislation.
Another critical aspect overlooked in the regulation of AI is the absence of robust terms of service agreements. These agreements, often accepted without thorough review by users, can contain clauses that enable unethical or illegal practices by AI companies. By exploiting these agreements, businesses can circumvent industry standards and best practices, potentially leading to harmful outcomes for users.
To avoid repeating past mistakes, it is essential to learn from the challenges encountered over the last two decades. Self-regulation within the tech industry has proven inadequate, and granting broad immunity to profit-driven corporations incentivizes behaviors that prioritize expansion over societal well-being. Establishing legal frameworks that limit Section 230 immunity and enforce platform compliance standards is crucial. These frameworks should include mechanisms for content moderation, robust reporting procedures, and swift responses to address concerns. Additionally, relying on terms of service agreements as a means to bypass industry regulations should not be permissible.
In conclusion, while President Biden’s AI regulation efforts have been lauded for balancing innovation with public interest protection, the inclusion of legal safeguards and enforcement mechanisms is imperative to ensure accountability and deter harmful practices in the AI sector. Implementing stringent rules aligned with the executive order could help mitigate risks associated with AI technologies and prevent the exploitation of users’ rights and safety.
