Welcome to Tech Beat, your daily read on what matters in technology.
Alexandre LeBrun, the CEO of AMI Labs — the world model startup backed by Yann LeCun — is pushing back on the industry's favorite buzzwords. LeBrun refuses to call his AI either AGI or superintelligence, a striking stance when nearly every major lab is racing to claim that finish line. It signals a growing tension between marketing ambition and engineering honesty.
Shifting to China's AI ambitions, Moonshot's upcoming Kimi K three model is expected to carry somewhere between two trillion and three trillion parameters, which would make it the largest open AI model to emerge from China. The Financial Times reports it could close the gap with Anthropic's Opus four point eight, a sign that the competitive pressure on Western labs is not slowing down.
And if you use Zoom on Windows, go update it now. The company patched a critical improper input validation flaw that could have allowed remote account takeover, alongside three additional high-severity vulnerabilities. All were found internally, with no evidence of active exploitation — but the window between patch and attack is rarely very wide.
Keep surfing. Tech Beat out.
