This past week, prominent figures from Silicon Valley, including the White House’s AI and crypto advisor David Sacks and Jason Kwon, the Chief Strategy Officer at OpenAI, stirred up considerable debate with their remarks regarding organizations that advocate for AI safety. In different instances, both individuals suggested that certain proponents of AI safety might not be as altruistic as they seem, implying ulterior motives either for personal gain or as agents of wealthy benefactors.
TechCrunch reached out to various AI safety groups, who argued that the accusations from Sacks and OpenAI reflect a broader trend in Silicon Valley aimed at silencing critics. This isn’t the first instance of such intimidation; in 2024, there were rumors propagated by venture capital firms that a California AI safety bill, known as SB 1047, would lead to severe consequences for startup founders. Despite being debunked by the Brookings Institution, who labeled it a misrepresentation, the bill was ultimately vetoed by Governor Gavin Newsom.
Recent controversies and accusations
Regardless of the intentions behind their statements, the comments from Sacks and Kwon have clearly caused apprehension among many advocates for AI safety. Numerous leaders from nonprofit organizations contacted by TechCrunch requested anonymity, expressing concerns about potential repercussions for their groups.
David Sacks’ critical view of Anthropic
This week, Sacks posted on social media platform X, criticizing the AI lab Anthropic. He accused them of using fear tactics to push through legislation that would primarily benefit their interests while sidelining smaller startups. Notably, Anthropic had been a vocal supporter of California’s Senate Bill 53, a law that mandates safety reporting for large AI companies, recently enacted.
Sacks’ remarks were prompted by a widely circulated essay from Anthropic co-founder Jack Clark, who shared his concerns about AI’s potential to contribute to unemployment and societal risks during a speech at the Curve AI Safety Conference. While many in the audience interpreted Clark’s words as an honest reflection of a technologist’s worries, Sacks perceived them quite differently, framing Anthropic’s discourse as a form of regulatory manipulation.
OpenAI’s legal actions against safety advocates
In a related development, Jason Kwon of OpenAI elaborated on the company’s decision to issue subpoenas to several AI safety nonprofits, including Encode, which promotes responsible AI policy. Kwon cited concerns stemming from a lawsuit filed by Elon Musk against OpenAI, claiming the company strayed from its initial nonprofit mission. This legal action raised suspicions about the motivations behind the criticisms directed at OpenAI’s restructuring.
Transparency concerns and industry reactions
Kwon expressed that the situation raised questions about the funding sources behind these organizations and whether there was coordinated opposition against OpenAI. Reports from NBC News revealed that OpenAI’s subpoenas requested communications related to Musk and Meta CEO Mark Zuckerberg, as well as inquiries into Encode’s support for SB 53.
Some AI safety advocates have noted a noticeable divide within OpenAI itself. While its safety researchers frequently publish warnings about the risks associated with AI systems, the company’s policy team has actively lobbied against SB 53, preferring a uniform regulatory approach at the federal level.
Joshua Achiam, OpenAI’s head of mission alignment, commented on the company’s actions, hinting at his discomfort with the approach taken. He remarked on social media that the current direction did not seem advantageous for the organization, hinting at internal dissent regarding the strategy.
The broader implications of AI safety movements
Brendan Steinhauser, CEO of the nonprofit Alliance for Secure AI, which has not been subpoenaed, claimed that OpenAI’s perception of its critics as part of a conspiracy orchestrated by Musk is misguided. He emphasized that the AI safety community is increasingly critical of the safety protocols—or lack thereof—employed by OpenAI and others.
Steinhauser pointed out that the actions taken by OpenAI appear to aim at stifling dissent and intimidating other nonprofits from voicing their concerns. Meanwhile, Sriram Krishnan, a senior policy advisor for AI at the White House, also weighed in, suggesting that AI safety advocates are disconnected from the realities faced by those using and adopting AI technologies in everyday scenarios.
Public sentiment appears to be shifting, as a recent survey indicated that nearly half of Americans express more worry than enthusiasm regarding AI. Many voters prioritize concerns over job displacement and deepfakes over catastrophic risks, which aligns more closely with the focus of the AI safety movement.
This evolving landscape raises critical questions about the balance between ensuring safety in AI development and fostering rapid growth in the industry, a concern that resonates deeply in Silicon Valley. As the AI safety movement continues to gain traction, it may provoke further resistance from industry leaders, indicating its growing influence.
