Key facts
- Frontier AI companies are seeking antitrust waivers and liability protections.
- Treasury Secretary Scott Bessent opposes liability exemptions for AI creators.
- Anthropic reportedly plans a $2 trillion IPO in October.
- Section 230 of the Communications Act provided internet platforms with liability protection.
- The University of Chicago barred AI from core social sciences courses.
Frontier artificial intelligence companies are seeking legislative protections, drawing parallels to the internet's early days and the creation of Section 230 of the Communications Act. Anthropic co-founder Dario Amodei has published an essay requesting antitrust waivers, arguing that AI labs need to slow down development together. However, Treasury Secretary Scott Bessent has opposed granting liability exemptions, stating that creators should be held accountable for the AI systems they build and generate.
In 1995, Congress grappled with the risks of the nascent internet, leading to Section 230, which shielded online platforms from liability for user-generated content. This provision, intended to encourage self-regulation, ultimately became a foundational element of the social media era, allowing platforms to grow without answering for the content they amplified. Today, AI labs are seeking similar protections, including antitrust waivers and immunity from lawsuits over AI-generated content.
Recent pushback has come from various officials. Jonathan Kanter, President Biden's antitrust chief, told CNBC that frontier labs can implement safety standards without needing an antitrust waiver. David Sacks, a former AI czar under President Trump, asserted that companies are responsible for making their products safe, and if they cannot control them, they should not proceed with development.
The article argues against embedding industry-friendly rules into statute during AI's early, poorly understood phase, citing the longevity of Section 230, which has been amended only once in 30 years. It suggests that a liability exemption or antitrust waiver could become entrenched, benefiting current market leaders. The author posits that AI labs asking for permission to self-coordinate on safety are also seeking to coordinate their market positions. The piece concludes that accountability is the price of profit and that AI labs capable of slowing down can do so without government permission.