In short
- Andy Konwinski, who co-founded Databricks and Perplexity AI, said this week that powering AI is a security risk in and of itself.
- The talk followed Open Frontier, a working meeting of about 100 researchers in San Francisco on June 30.
- Turing Award winner Yann LeCun responded directly to X, comparing today’s closed AI moment to “medieval obscurantism like the Ottoman Empire banning the use of printing presses for 200 years.”
Perplexity AI and Databricks co-founder Andy Konwinski thinks the AI security debate is flawed: It’s being used to power, not harm. Earlier this week, he published an article detailing his case, with Anthropic as the star witness.
The story they build begins Anthropic decision was changed in 48 hours. When Anthropic launched Claude Fable 5 on June 9, a paragraph included in its 319-page prospectus revealed that the prototype would quietly mock its responses to anyone who thought it was training a competitive AI.
Researchers found. The Internet did not take it well.
Anthropic is back, but for Konwinski that doesn’t make a difference in assessing the big picture. “The problem isn’t that Anthropic made the wrong decision,” he wrote. “The problem is that they thought the decision was theirs.”
His storytitled “Consolidation of power in AI is a risk, not a solution,” followed Open Frontier, a working conference he convened through his non-profit Laude Institute at San Francisco’s Exploratorium on June 30. About 100 researchers showed up.
UC Berkeley Dean Jennifer Chayes, who runs the College of Computing, Data Science, and Society, told the funding group that Berkeley researchers “are all building on Chinese models because we don’t have open Western borders”—and that the security messages from OpenAI and Anthropic ahead of their IPOs were “a very dangerous project.”
Konwinski’s argument is that finding a middle ground does not reduce risk; it creates another. AI is the foundation – in the same category as railways, electricity, and the Internet. Those technologies also connected all the people who managed the event. The same is coming to AI. Instead: collaborative research with a border that allows top researchers to reach the border without needing permission from a private laboratory to do so.
LeCun: It’s the Ottoman Empire that banned the printing press
Yann LeCun, who was Meta’s chief scientist, responded to Konwinski’s article on X with no clarity. “I’ve been spreading the same message for years,” he responded to Konwinski’s post. “Too much power in AI and the desire to control it is the greatest danger of AI.”
He also had a comparable record of preparation. “It’s a type of medieval obscurantism similar to the Ottoman Empire’s 200-year ban on the use of the printing press, in part to control doctrine, and to protect the guild of calligraphers and scribes,” LeCun wrote.
LeCun’s final prediction: “Architecture wants to open up. Foundation models are becoming important and will undoubtedly be sold. For the long term, the money is in the utilities.”
LeCun left Meta at the end of 2025 and started AMI Labs in Paris and $ 1.03 billion in seed money in March 2026-his answer to this question. The company operates on international brands with its JEPA design, plans to open its own research, and does not have expected sales for many years.
Daily Debrief A letter
Start each day with top stories right here, including originals, podcasts, videos and more.
