Elon Musk thinks Grok 5 might actually reach AGI – and that’s something we should all be worried about (www.techradar.com)

Elon Musk tweeted that he now believes xAI’s upcoming Grok 5 “has a chance of reaching AGI,” suggesting the model — slated for release by year’s end — could produce human‑level general intelligence far sooner than many expected. That shift is notable because Musk previously backed calls for a moratorium on advanced AI research; his new assessment, echoed by high‑profile worries from figures like Geoffrey Hinton, thrusts xAI’s roadmap into the center of debates about timelines, governance, and whether a single company could control transformative capabilities. Experts urge caution: while Grok 5 may markedly improve digital reasoning — conducting complex web‑based research and delivering far better answers — researchers like Himanshu Tyagi doubt it will immediately solve genuinely novel scientific problems (e.g., discovering new proteins). The core technical implication is a convergence toward systems that excel across many online tasks without necessarily achieving the full breadth of human cognition. That “AGI‑light” jump still increases risks: opaque, speed‑prioritizing models can amplify safety, security, and control challenges. The development highlights the tradeoffs between proprietary, fast‑moving efforts and open‑source alternatives that prioritize transparency and community oversight.