Open-weight AI models are catching up to the frontier. The safety gap remains.
The report highlights that GLM-5.2, a Chinese open-weight AI model, demonstrates strong capabilities in cyber and bio tasks, outpacing models like Claude Opus 4.7 in certain scenarios, raising concerns about the risks of uncontrolled AI use. While Z.ai implemented safety measures for its hosted API, these protections are ineffective when models are run on personal hardware, where attackers can bypass safeguards. The debate centers on how to manage risks with open-weight models, advocating for pre-training data filtering and closed-model safeguards, but acknowledging the challenges of implementing such measures. Chinese companies like Z.ai have not provided detailed safety frameworks or risk assessments for GLM-5.2, prompting discussions on the need for rigorous evaluation and control of advanced
More from Techcrunch
Microsoft kills off unsuccessful AI features while merging its separate Copilot apps
Microsoft is simplifying Copilot by combining its consumer and business apps, and dropping AI-generated podcasts, Group Chats, Deep Research, and its Mico character.
techcrunch.com··Mads Kristian Nylund·AI ToolsAI InfrastructureAI Governance
Nvidia’s new $500B plan is risky but brilliant, especially for aging GPUs
Nvidia has a plan to make sure its GPUs won't lose value. It wants to convince a new crop of financiers to keep lending for AI buildouts.
techcrunch.com··Mads Kristian Nylund·AI DevelopmentAI InfrastructureAI Governance
Apple in talks to pay publishers to provide Siri with current news: report
The tech giant has considered a nine-figure budget for the payments, according to the WSJ.
techcrunch.com··Mads Kristian Nylund·AI ToolsAI EthicsAI Platforms