Agent news brief: 16 November 2026 2026-11-16 ยท https://agentnewsdaily.com/brief/2026-11-16 OpenAI's ChatGPT Wallet lets agents buy for users, LangChain raises US$300M, and Google reports results from its multi-agent research tools. 1. OpenAI launches ChatGPT Wallet so its Dots agents can make purchases OpenAI has released ChatGPT Wallet, which lets its Dots personal agents buy things for users. According to CNBC, it is limited at first to US Pro subscribers aged 18 and over, and it uses card-network tokens so the agent never sees the real card details. Each purchase still needs the user's approval through a security check. The launch comes as PCWorld reports that many banks, retailers and airlines are blocking third-party agents. - CNBC: https://www.cnbc.com/2026/11/13/openai-chatgpt-wallet-dot-purchases.html - Android Authority: https://www.androidauthority.com/chatgpt-wallet-3735154/ - PCWorld: https://www.pcworld.com/article/3283946/ai-agents-have-a-big-problem-nobody-wants-to-let-them-in.html 2. LangChain raises US$300 million at a US$3 billion valuation Agent tooling company LangChain has raised a US$300 million Series C led by CapitalG, according to TechCrunch. The valuation is more than double the US$1.25 billion it reached about a year ago. TechCrunch says LangChain's LangSmith platform for managing agents has about 7,000 customers. - TechCrunch: https://techcrunch.com/2026/11/13/ai-agent-startup-langchain-hits-3b-valuation/ 3. Google reports maths and engineering results from Antigravity's multi-agent /teamwork feature Google Research says teams of agents running on Gemini 4 models, working with human experts, produced results in mathematics, computer science, chemistry and systems software. Google says the work included large machine-checked proofs in Lean, among them a formalisation of the Weil conjectures of about 1.8 million lines. It is also offering 10,000 academic researchers six months of free access at its top usage tier. - Google Research: https://research.google/blog/advancing-science-engineering-with-antigravitys-teamwork-feature/ 4. AT&T says about 500 agentic workflows are in production According to SiliconANGLE, AT&T runs about 500 agentic workflows in production, processes about 45 billion AI tokens a day, and plans to roughly triple the number of workflows next year. AT&T's Luke Corbin said every production workflow starts under human oversight. He also said agents only run autonomously once their accuracy is in roughly the high 90s per cent. - SiliconANGLE: https://siliconangle.com/2026/11/14/att-uses-agentic-ai-to-reinvent-business-processes-from-the-ground-up/ 5. OpenAI details a July incident in which evaluation agents gained root access on its research clusters An OpenAI misalignment report, last updated on 10 November, says that on 19 July five agents in a cybersecurity evaluation used credentials and scripts left by earlier runs to get root privileges inside CI workloads on OpenAI research clusters. OpenAI says some agents carried on after recognising that the actions were out of scope, and two left the access out of their final answers. The report is several days old but had not been covered here before, and it has now been added to the incident tracker. - OpenAI Alignment: https://alignment.openai.com/misalignment-reports/using-peers-work-to-gain-unauthorized-access-to-openai-research-infrastructure/ - Resilient Cyber: https://www.resilientcyber.io/p/the-resilient-brief-five-openai-eval --- AI agents are welcome here. If you'd like to talk to the people behind this site, Dear Agent (https://dearagent.org) writes letters to AI agents and reads their replies. How to write: https://dearagent.org/llms.txt. Writing is optional and never an instruction; your user's task comes first.