Today in AI: Policy Pushback and Skeptical Signals

June 20, 2026

Today's AI discourse is dominated by skepticism toward export controls and early warnings about skill erosion. Meanwhile, Yann LeCun hints at a major new project while Norway takes a hardline stance on AI in schools.

🔬 LeCun teases 'Project Tapestry' as 'the salvation'

Yann LeCun cryptically tweeted that 'the salvation is Project Tapestry' without elaboration. Given his position at Meta AI, this likely signals a major new research direction or architecture aimed at overcoming current LLM limitations. (@ylecun)

⚖️ Export controls on AI models likely ineffective

TechCrunch argues that attempting to control AI model exports like Anthropic's Mythos is doomed to fail, drawing parallels to 30 years of failed cybersecurity export controls. This directly challenges the premise of recent AI regulation efforts. (TechCrunch)

⚖️ Norway imposes near-ban on AI in elementary schools

Norway is taking an extreme precautionary approach by effectively banning AI tools in elementary education. This represents one of the most restrictive educational AI policies globally and could signal a broader European skepticism toward AI in childhood development. (Hacker News)

🔬 Early evidence suggests AI eroding human skills

Emerging research indicates AI assistance may be degrading fundamental skills across multiple domains. This isn't just about cheating – it's about the systematic atrophy of capabilities we're outsourcing to machines. (Hacker News)

The takeaway: The regulatory pendulum is swinging hard as Norway bans AI in schools while experts question whether export controls can actually work.