# Why we need to ACCELERATE AI research, not slow it down  ## 1. Offense and defense share the same engine The models that find exploits are the models that patch them, write detectors, and investigate in production. The solution to AI-powered cyberattacks is AI-powered cyberdefense. > Do you really want to slow down and lag in cyber-defense? ## 2. Safety research needs better models Freeze the frontier, and safety research freezes with it. Think of AI interpretability as trying to infer a human's behavior from brain scans. Exactly what a more powerful AI would improve. As [**Yann LeCun**](https://x.com/ylecun/status/2100144656726073500?s=46&t=qkQEe9nFW0IQT6AdLZv4eg) phrased it: > Progress in airplane design have made them safer and safer. Airplanes would have been safer earlier with accelerated progress. The same is true for AI: better AI is safer AI. ## 3. AI is already saving lives In medicine, AI already helps with clinical decision support, reduces diagnostic errors, and shortens the path from molecules to treatments. On our roads, autonomous cars are already preventing deadly accidents. > Slowing the frontier delays lives saved, not just research papers published. ## 4. Safety needs competition too Open, fast frontier work makes potential failures public earlier and forces competing safety approaches instead of relying on one stack. > A limited number of labs without competition due to regulatory capture is a recipe for failure.