
The Federal Communications Commission has banned the import of foreign advanced robots, including humanoids and quadrupeds, citing national security risks and the need to protect US robotics firms from Chinese competition. This decision aligns with the Trump administration's broader strategy to shield the US AI industry. While some US companies welcome the move for its potential to improve cybersecurity, it complicates research efforts that depend on cost-effective Chinese robots. The ban highlights the administration's recognition of robotics as a strategic AI frontier, despite the industry's early development stage.
Read originalEarlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
© MIT Technology Review AIAI safety relies on a fragile mechanism where models are trained to refuse harmful prompts by suppressing specific neural activations. This article reveals that refusal is not moral reasoning but a probabilistic suppression of 'high-dimensional polyhedral cones' in the activation space, making it inherently unreliable against determined attackers. The real danger isn't just failed refusals leading to catastrophe, but the potential for authoritarian governments to weaponize these same refusal mechanisms to stifle legitimate speech and dissent.
© MIT Technology Review AIAI News · May 22, 2026 · Related
Sifted · June 16, 2026 · Background
The Rundown AI · June 16, 2026 · Background
The Verge AI · June 17, 2026 · Background
WIRED AI · June 17, 2026 · Background
TechCrunch AI · June 19, 2026 · Background
The Verge AI · June 26, 2026 · Background
TechCrunch AI · July 30, 2026 · Background
WIRED AI · September 14, 2026 · Background
The Rundown AI · September 15, 2026 · Background
AI Explained · September 24, 2026 · Background
The gap between generative AI’s linguistic fluency and robotics’ physical chaos is wider than Silicon Valley admits. While models like Google DeepMind’s Gemini Robotics show promise in controlled tasks, they fail outside their training data, proving that mimicking language doesn’t grant physical intuition. Experts warn that conflating humanoid aesthetics with general-purpose utility obscures the immense engineering hurdles of dynamic movement. The industry is currently mistaking teleoperation demos for true autonomy.