Build, discover, and monetize AI agents.

Anthropic accuses Alibaba-linked operators of mass extraction of Claude’s agentic capabilities

What changed: Anthropic told U.S. officials that operators linked to Alibaba’s Qwen lab used tens of thousands of fraudulent accounts and millions of exchanges to distill and extract Claude model capabilities — including software-engineering and agentic reasoning features.

Why it matters: If true, this is a reminder that proprietary agent reasoning and orchestration logic can be targeted at scale; buyers and builders need stricter API controls, usage monitoring, and vendor risk assessments for any agent that holds competitive or regulated knowledge.

Try/watch: Treat vendor model provenance as a procurement risk: require transparency on access controls, rate limits, and anti-abuse measures, and monitor regulatory or legal fallout that could affect cross-border agent deployments.

Nokia and Google Cloud embed Gemini agents into network operations software

What changed: Nokia expanded its Assurance Center by integrating Google Cloud’s Gemini foundation models into a coordinated six-agent system for network triage, KPI selection, anomaly reasoning, remediation and dashboards, with a staged SaaS rollout and initial agents already live.

Why it matters: Operators and service providers can now buy a prebuilt multi-agent stack for routine network ops that claims large reductions in troubleshooting time, but you should expect staged adoption with human sign-off on high-risk actions.

Try/watch: Pilot the event-triage and dashboard agents first, enforce "glass-box" approvals for any automated remediation, and validate claimed MTTR gains against your real telemetry to guard against risky automation drift.

Board adds Supply Chain and Merchandiser agents to its ‘Agentic Continuous Planning’ platform

What changed: Board released domain-specific Supply Chain and Merchandiser agents that run inside its planning environment to connect forecasting, scenarios, and operational workflows so decisions can be continuously evaluated and acted on.

Why it matters: Buyers in retail and operations can move from episodic forecasting to continuous, governed decision loops where agents suggest and execute trade-offs that align supply, inventory, and financial objectives — provided the data and audit trails are managed.

Try/watch: Run a single-domain pilot (e.g., category-level merchandising) with clear audit and rollback procedures, and insist on explainable recommendations so finance and operations can validate agent-suggested trade-offs.