White House Advisor Tells OpenAI and Anthropic to Take Liability for AI Safety
White House tech advisor David Sacks criticized OpenAI and Anthropic, arguing that AI companies should take full responsibility for building safe products rather than seeking government intervention or antitrust exemptions to slow down development. He echoed views that firms should face civil or criminal liability if their AI models cause severe harm, such as destructive cyberattacks. This underscores a policy debate favoring direct product liability and market mechanisms over regulatory slowdowns or industry-wide collusive safety pauses. It highlights tension between major AI labs advocating for government-sanctioned coordination and policymakers urging companies to internalize the legal and financial risks of unsafe releases. Sacks specifically rebutted Anthropic CEO Dario Amodei's call for antitrust waivers that would allow AI rivals to coordinate a slowdown in model development. He agreed with former FTC Chair Lina Khan that existing product liability frameworks apply to AI, asserting that market forces already penalize unpredictable or unauthorized model behavior.
## BACKGROUND
As generative AI models rapidly advance in capabilities, concerns have grown over risks such as automated cyberattacks and misinformation. Some AI leaders advocate for legal mechanisms and government oversight to pause or align development across the industry, whereas critics argue such proposals may be attempts to avoid product liability or create regulatory barriers against newer competitors.