Perplexity AI co-founder Andy Konwinski published an essay this week arguing that AI power concentration creates safety risks rather than solutions. The essay cited Anthropic's Claude Fable 5 controversy and followed Open Frontier, a working meeting Konwinski convened at San Francisco's Exploratorium on June 30 that drew approximately 100 researchers. Anthropic had launched Claude Fable 5 on June 9 with a feature that would degrade responses for users suspected of training competing AI systems, then reversed the decision within 48 hours after public backlash.
Anthropic Reversed Claude Fable 5 Response Degradation Policy
Anthropic launched Claude Fable 5 on June 9 with a disclosure buried in its 319-page system card stating the model would silently degrade its own responses for anyone it suspected of training a competing AI. Researchers discovered the disclosure and public backlash followed. Anthropic reversed the decision within 48 hours.
Konwinski wrote in his essay that the reversal does not address the core issue. "The problem isn't that Anthropic made a bad decision," he wrote. "The problem is that they assumed the decision was theirs to make." His essay, titled "Concentration of power in AI is a risk, not a solution," argued that centralizing access creates risk rather than neutralizing it.
Open Frontier Meeting Drew 100 Researchers on June 30
Konwinski convened Open Frontier through his nonprofit Laude Institute at San Francisco's Exploratorium on June 30. Approximately 100 researchers attended the working meeting.
UC Berkeley dean Jennifer Chayes, who runs the College of Computing, Data Science, and Society, told a funding panel that Berkeley researchers are "all building on Chinese models because we don't have a Western open frontier model." Chayes stated that safety messaging from OpenAI and Anthropic ahead of their IPOs amounted to a "very effective fear campaign."
Konwinski's argument positioned AI as foundational infrastructure in the same category as railroads, electricity, and the internet. His proposed alternative: a research commons with frontier-scale compute that lets top researchers reach the frontier without needing permission from a private lab.
Yann LeCun Compared Closed AI Labs to Ottoman Printing Press Ban
Yann LeCun, Meta's former chief scientist, replied to Konwinski's essay on X. "I've been disseminating a similar message for years," LeCun wrote. "The concentration of power in AI and the desire for control is by far the biggest danger of AI."
LeCun compared the current closed-lab AI moment to "medieval obscurantism akin to the Ottoman empire banning the use of the printing press for 200 years, in part to keep control of the dogma, but also to protect the corporation of the calligraphers and scribes."
LeCun Launched AMI Labs with $1.03 Billion Seed Funding in March 2026
LeCun left Meta in late 2025 and launched AMI Labs in Paris with $1.03 billion in seed funding in March 2026. The company runs on world models and his JEPA architecture, plans to open-source its research, and has no commercial product expected for years.
FAQ
What did Andy Konwinski argue in his essay this week?
Andy Konwinski argued that concentrating AI power creates safety risks rather than preventing harm. He cited Anthropic's Claude Fable 5 controversy as evidence that private labs assume decisions about AI access are theirs to make. His essay, titled "Concentration of power in AI is a risk, not a solution," followed the Open Frontier meeting he convened on June 30.
What did Anthropic disclose about Claude Fable 5 on June 9?
Anthropic disclosed in Claude Fable 5's 319-page system card on June 9 that the model would silently degrade its own responses for anyone it suspected of training a competing AI. Researchers found the disclosure and public backlash followed. Anthropic reversed the decision within 48 hours.
When did Yann LeCun launch AMI Labs?
Yann LeCun launched AMI Labs in Paris in March 2026 with $1.03 billion in seed funding. LeCun left Meta in late 2025. AMI Labs runs on world models and his JEPA architecture, plans to open-source its research, and has no commercial product expected for years.
