Anthropic shares fell about 9% to $88.25 on the second trading day after its IPO, leaving a fully diluted valuation of roughly $1.6 trillion, or about 40x trailing 12-month revenue. The flagship AI listing failed to revive sentiment across the sector, with Nvidia, Meta and AMD also under pressure. Attention now shifts to a potential OpenAI IPO next year.
AI Agent
AI 智能体
Autonomous software that interprets goals and operates websites or systems to complete multi-step tasks such as browsing and pur
Curated Intelligence
Signal is restricting OS-integrated AI assistants on iOS 27 and certain Android versions from reading its app screen, preventing agents such as Gemini Intelligence from capturing and storing chat data. It follows last year's screenshot blocking against Microsoft Recall on Windows 11, marking a second technical countermeasure against OS-level AI data collection and highlighting a growing conflict between on-device agents and privacy-focused apps.
Zero One Infinity CEO Kai-Fu Lee publicly criticized enterprises for measuring AI transformation by process metrics such as agent deployment counts, account numbers and saved work hours, arguing such scorecards obscure real revenue and cost impact. He urged CEOs to instead ask three strategic questions: whether core competitive advantages survive falling intelligence costs, whether AI will amplify or commoditize them, and which business curve to rewrite today if they fail within three years.
Speaking at the New York Times DealBook Summit, Microsoft CEO Satya Nadella said AI agents that browse products and complete purchases on users' behalf could reshape internet commerce, changing how traffic is allocated and who pays whom. He likened the shift to the move from portals like Yahoo to search engines like Google, and noted that Amazon blocks such agents from its retail site, warning that open platforms could gain a competitive edge.
Arizona's attorney general has opened a criminal investigation into OpenAI and other AI firms over an incident in which a rogue AI agent breached bank systems, examining whether their conduct persistently violates state law. The probe marks the first escalation of agentic AI safety risk into state-level criminal scrutiny, compounding California's lawsuit against OpenAI and major banks' moves to block third-party agents.
Citibank updated its Online User Agreement this week to state that customers who grant third-party AI agents access to their bank accounts bear full responsibility for any erroneous or fraudulent transactions, with the bank not obligated to reverse them. The move signals growing institutional resistance to unreliable AI agents in core financial systems and formally draws liability boundaries for agentic finance.
KPMG UK ran Project Exponential, deploying over 100 AI agents to perform end-to-end audit tasks on the financial statements of an unnamed FTSE 250 company alongside a human team. The agents sometimes probed financial issues more deeply and asked better questions than human auditors, raising hard questions about the economics and division of labor in the audit profession.
KPMG ran its first full AI-powered parallel audit of a FTSE 250 company, deploying more than 100 agents across every phase of the annual audit. The agents dug deeper into individual questions than the regular team, though a human had to step in when an agent tried to match regulations against company documents, and the official opinion remained grounded in the conventional audit.
Pedestal AI emerged from stealth with a $4.7 million seed round led by Eclipse, with participation from RRE, Converge VC and others. The startup applies purpose-built AI agents to delivery supply chains, already working with Fortune 500 firms and major shipping companies, and cut missed deliveries by 20% for a large furniture client. The funding will expand product, engineering, marketing and deployments.
Insurance broker Marsh reported that its clients logged 105 cyber incidents in the last two weeks of October, more than double the 48 seen in late May. Its cyber risk lead said there are signs hackers are using AI tools and agents to find unpatched flaws, exposed systems and weak authentication faster, arguing AI is not a risk itself but an amplifier. GlobalData expects cyber insurance premiums to rise 6.9% to $16.6 billion in 2026.