Brian Merchant delivers a stinging indictment of California's recent AI legislation, arguing that the state's political machinery has been so thoroughly captured by Silicon Valley that the resulting laws are not just ineffective, but actively harmful to public safety. While headlines celebrate "sweeping" new regulations, Merchant exposes a reality where meaningful worker protections and child safety measures were vetoed, leaving behind only performative compliance that shields tech giants from accountability.
The Illusion of Reform
Merchant begins by dismantling the narrative of progress, noting that Governor Gavin Newsom signed bills that were largely toothless while vetoing those that posed actual risks to the tech industry's bottom line. "With two exceptions, things broke just the way I expected them to: Newsom signed the toothless bills and vetoed those the tech industry took issue with," Merchant writes. This framing is crucial because it shifts the focus from legislative output to legislative intent, revealing a system where the appearance of action is prioritized over substantive change.
The author highlights a rare victory: the passage of AB 325, which targets algorithmic price-fixing used by landlords and retailers. "We're thrilled that California will make abundantly clear that whether or not you shake hands on a back room deal or use an algorithm to artificially increase prices, California will hold you accountable," says Samantha Gordon of TechEquity, a group Merchant notes backed the law. However, Merchant quickly pivots to the broader failure, arguing that this single win is overshadowed by the systematic dismantling of stronger safeguards.
The verdict is in, and, surprise, it (mostly) does. Silicon Valley's institutional political power has, for now, become all but insurmountable.
Merchant's analysis of the vetoes is particularly sharp. He details how the administration blocked the "No Robo Bosses Act" (SB 7), a bill that would have prevented employers from using AI to fire workers without human review. "Silicon Valley lobbied for the right for its AI to fire you without a human manager in the loop, and won," Merchant observes. This is a stark reminder that the push for automation often comes at the direct expense of labor rights, a dynamic that the administration chose to endorse rather than challenge.
The Danger of Performative Safety
The most damning section of the piece concerns the veto of the LEAD Act (AB 1064), which would have required AI chatbots to be safe for children before market release. Merchant critiques the administration's justification, which claimed the bill was too broad and might inadvertently ban all AI for minors. "Heaven forbid. What if, in the process of trying to ban AI products that quite actually encourage children to kill and harm themselves, we wind up banning chatbots that help children cheat on their homework, diminish their propensity for critical thought, and lead to the development of other forms of AI psychosis?" Merchant sarcastically asks, exposing the absurdity of prioritizing market access over child safety.
Instead, the administration signed the Transparency in Frontier Artificial Intelligence Act (SB 53), a law Merchant describes as "almost comically pointless." The bill requires companies to self-report catastrophic risks and post safety protocols, with a maximum fine of $1 million—a sum Merchant notes is "less than the wire transfer fee from OpenAI's latest SoftBank loan distribution." This critique lands hard because it highlights the futility of relying on the very entities creating the risks to police themselves.
Critics might argue that even weak regulations are a starting point for future improvements, but Merchant counters that these laws create a false sense of security that actually hinders further progress. "The appearance of having passed meaningful laws around AI risks sapping the political will to meaningfully tackle actual AI social and labor issues," he argues. By passing laws that do nothing, the state effectively blocks the path for more robust legislation.
The Cost of Capture
The piece concludes by broadening the scope from California to the national stage, suggesting that if a liberal state like California cannot resist corporate pressure, the rest of the country stands little chance. "We have to start thinking about what it means that at least for now, US citizens effectively have no meaningful democratic input into how technology shapes our workplaces, institutions, and civil society," Merchant writes. This is a sobering assessment of the current political landscape, where the gap between public interest and corporate power has become unbridgeable through traditional legislative channels.
Merchant does offer a glimmer of hope, pointing to grassroots movements and the potential for legislative overrides, but the overarching tone is one of urgent warning. The argument is that the capture of political institutions is not a future threat but a present reality, and the recent legislative session in California serves as the definitive proof.
Silicon Valley's capture of our political institutions is all but complete.
Bottom Line
Merchant's strongest contribution is his ability to dissect the specific mechanics of how lobbying translates into policy failure, moving beyond vague accusations of corruption to show exactly which bills died and why. The piece's greatest vulnerability is its reliance on a binary view of legislative success, potentially underestimating the long-term value of establishing even weak regulatory frameworks. However, the evidence that meaningful protections were actively vetoed while performative ones were signed makes a compelling case that the current trajectory is dangerously skewed toward corporate interests.