Perplexity Co-Founder: AI Safety Is an Excuse to Lock Down Frontier
Andy Konwinski used Anthropic's Fable 5 debacle as exhibit A in the case against letting a handful of private labs govern who gets to do AI research.
Andy Konwinski used Anthropic's Fable 5 debacle as exhibit A in the case against letting a handful of private labs govern who gets to do AI research.
Read Full Story at Decrypt →Why This Matters
The debate over AI safety has rapidly evolved from theoretical caution into a battleground over control of the technology's future. Konwinski’s argument reframes the issue as a power struggle, where calls for regulation may disguise efforts to consolidate AI development under a select few corporate entities. This challenges the narrative that safety advocates are purely motivated by public good, instead highlighting how competing visions of oversight could reshape the entire AI ecosystem.
Background Context
The Fable 5 incident exposed the fragility of AI systems when pushed beyond their training parameters, but it also became a lightning rod for criticism of Anthropic’s handling of the debacle. Konwinski’s critique taps into a long-standing tension between open research and proprietary control in AI, where startups and academics increasingly clash with well-funded labs over who sets the rules. This divide mirrors historical patterns in tech, where early ideals of democratization often collide with commercial and geopolitical realities.
What Happens Next
Expect intensified lobbying as private labs push for frameworks that favor their interests, while policymakers grapple with how to balance innovation against accountability. The open question is whether grassroots advocates or institutional bodies will set the terms of debate, potentially leading to fragmented regulations that disadvantage smaller players. Watch for how Congress and international bodies like the EU AI Act adapt—or resist—calls to centralize AI governance under a handful of gatekeepers.
Bigger Picture
This clash underscores a broader reckoning in tech: where safety becomes a proxy for control, and where the promise of open innovation collides with the realities of capital and regulation. It also reflects a global shift toward tech sovereignty, where nations and corporations alike seek to dominate AI not just for its economic potential, but for its strategic leverage. The outcome could redefine not just who builds AI, but who gets to decide what AI can—and cannot—do.
