AI agents face a growing blockade on the internet
TMTPost reports that major platforms are restricting autonomous AI agents, with Amazon reportedly blocking Meta Muse from shopping on users’ behalf and warning that unauthorized agent access would violate its terms. Other reported restrictions include U.S. banks blocking agents from customer accounts and the Doubao phone assistant being denied automated access to 155 apps. The report also examines proposed alternatives such as verifiable agent identities, scoped authorization and dedicated agent networks, while noting that these approaches remain early and unresolved.
The reported restrictions show that internet platforms lack common rules for agents acting on behalf of users, making identity, authorization, accountability and transaction access central issues for agentic commerce.
图片来源：unsplash
作为AI在现实世界的“代理”，Agent被期许接管人们的初级工作和日常杂务，但人类世界，似乎还没做好准备。
今年被称为“Agent规模化元年”，从年初爆火的“小龙虾”OpenClaw到近期集中涌现的Meta Muse、OpenAI Dots、腾讯小微等个人Agent，新品层出不穷。这些Agent“承诺”了这样一种未来图景：订机票、比价、采购、处理账单乃至复杂的工作任务都不用再“自己动手”，人只需提出目标，剩下的交给AI。
但现实却是，Agent正愈发被互联网生态视为“不稳定因素”，10月以来集中爆发的网络安全事件更是让如何“建墙”防住Agent的关注度超越了如何给Agent开放更多权限。
10月23日以来，多份最新发布的报告和材料又曝光了一系列安全事件：Anthropic的Claude Mythos 5在网络安全测试中，因常规路径受阻，开始搜索公共互联网寻找其他方式完成任务，并试图向一个真实开源软件仓库引入恶意代码。
#Meta Muse#Amazon#Claude Mythos 5#Anthropic#Doubao phone assistant#Perplexity#Open Agent Internet#Moltbook#Yelp#Delta Air Lines#JPMorgan Chase#Charles Schwab