OpenAI put its Agents API into public beta with a managed Codex harness that handles sessions, orchestration, context compaction, and recovery — plus a new finance workspace. By taking on the agent plumbing, OpenAI pushes the platform battle from raw models toward full agent runtimes, lowering the bar for teams shipping autonomous software.
A security-tuned model with frontier-level autonomous vulnerability discovery that Google says beats rival frontier models on the task. Defensive-AI capability is becoming a key differentiator among the big labs.
The AI-chip startup raised the two-tranche round just seven months after a $230M raise, betting on its HBM-free Asimov design with up to 2.3TB of LPDDR5X per die — a credible push against incumbent accelerators.
The legal-AI startup extended one of the fastest valuation climbs in enterprise AI. Capital is concentrating in AI that sits close to regulated, high-value professional workflows.
Meta is buying the multiplayer "AI teammate" startup to strengthen its business AI products — another sign platforms want collaborative, in-workflow agents over standalone chatbots.
The three unveiled an interoperability standard (KYA) aligning three previously competing protocols for verifying AI agents in payments. As agents start transacting, the industry is racing to define trust and identity rails.
Elon Musk's tunneling venture landed one of the month's largest rounds, a reminder that capital is flowing to physical-operations and infrastructure plays alongside pure software.
OpenAI says it made substantial fresh progress on another Millennium Prize math problem (Tech Startups) · Securin shipped its AI-native Preemptive Exposure Management platform to general availability (Help Net Security).