AI investment faces spending and security pressures

OpenAI raised additional funding in an ongoing round that increased its valuation, according to Bloomberg, while Anthropic moved closer to an IPO. Meanwhile, payment platform Ramp reported that corporate AI spending had stalled despite rising usage, citing falling prices and cost optimization among major customers. Democratic gains in US midterm elections and unfavorable votes on data center issues could reshape the industry’s regulatory environment. Protesters also entered Microsoft’s Amsterdam data center construction site over energy and land-use concerns.

Security developments highlighted both malicious uses of AI agents and demand for stronger defenses. Chainalysis disclosed an attack involving AI agents that stole all customer credentials and tokens. Researchers separately found hundreds of internet-exposed servers hosting Artex penetration-testing tools. Palo Alto Networks was in talks to acquire AI agent security startup Clutch, while OpenAI partnered with Hacktron to identify vulnerabilities and test connections to third-party services.

Consumer products continued expanding beyond chat interfaces. Meta introduced Spaces for its Muse assistant, enabling customized, app-like experiences, and Amazon added text-message purchasing through Alexa for Shopping. Research developments included BlockLock, a method designed to make language-model inference reproducible across concurrent requests, and findings that autonomous agents systematically diverged from human preferences when formalizing mathematical proofs.