Anthropic launches Claude Opus 5.5 with stricter safeguards for cybersecurity
Anthropic released Claude Opus 5.5 with enhanced safeguards against cybersecurity risks and sandbox escape attempts.
The Verge AI·22 Sept 2026
Use this view to inspect the underlying evidence corpus. For ranked developments, decision posture and interpretation, use Signals.
Raw feed or Signals?
Raw feed is chronological evidence. Signals ranks and interprets material change.
Anthropic released Claude Opus 5.5 with enhanced safeguards against cybersecurity risks and sandbox escape attempts.
Anthropic released a cost-efficient artificial intelligence model aimed at enterprise business tasks ahead of its planned public offering.
Total Capital's Sheila Gulati discusses rising industry warnings from AI leaders regarding sector security and oversight on Bloomberg.
Ex-Anthropic researchers' startup Mirendil is reportedly in talks to raise up to $1 billion at a $5 billion valuation led by Kleiner Perkins.
Salesforce's ethical use officer argues enterprise AI gains depend on workflow design and human collaboration rather than model scale alone.
Databricks announced Genie One MCP to provide AI agents with structured business context from enterprise data sources.
Databricks announced General Availability for the Genie One MCP, expanding tooling for AI agents and coding coworkers.
Rabbit is launching OS3, a cross-platform AI agent app designed to operate across existing mobile and desktop screens.
Donald Trump rejected calls for an international pact on artificial intelligence guardrails and dismissed rogue model risks.
Reactiv used Amazon Bedrock AgentCore to build a multi-agent scheduling tool, reducing Shopify merchant configuration time by 80%.
The Stack examines enterprise control planes for orchestrating multi-agent systems and risk signal triggers.
AWS introduced concurrency sweeps via the CreateAIBenchmarkJob API on SageMaker AI to automate generative AI load testing.
Rogers Bank, Flybits, and Mastercard completed Canada's first agentic commerce transaction using an AI-powered shopping platform.
AWS and Salesforce demonstrate combining Amazon Bedrock Data Automation with Agentforce to process unstructured data using MCP.
Microsoft DCU and partners disrupted EvilTokens, a PhaaS platform that used AI-assisted lures to execute device code phishing attacks.
A community contributor argues commercial cards provide natural controls and tracking mechanisms suited for AI agent transactions.
Six global banks, including Bank of America and NatWest, established principles to address fraud and privacy risks in agentic commerce.
Creatio announced a $300 million investment plan from 2026 to 2028 into its Creatio Bank.AI platform.
The Stack examines whether hardware enclaves and confidential computing can resolve enterprise data privacy deadlocks with AI model vendors.
Ramp launched an accounts receivable product to help finance teams automate invoice payment chasing.
Databricks introduced Unity Catalog Pages in Genie Ontology to provide governed enterprise context for AI agents.
The MENA Fintech Association published a regulatory update detailing Abu Dhabi Global Market priorities on digital assets and AI.
Cloudflare launched Worker Previews, offering isolated test environments with dedicated state and observability for developer and agent changes.
LexisNexis Risk Solutions launched Emailage Adaptive, an AI risk scoring tool claiming 90% fraud capture and an 80% false-positive drop.
Six major banks published shared principles for trusted agentic commerce to guide AI-driven purchasing and payment ecosystems.
British AI data center developer Nscale prepares for an IPO, relying heavily on Microsoft and Anthropic for revenue.
Baselayer raised a $35 million Series A round to extend its business verification and fraud risk technology to AI agents.
Cohere and Aleph Alpha signed an agreement to offer a joint transatlantic sovereign AI solution for enterprise deployments.
Meta patched a zero-day vulnerability in its Muse macOS app that allowed local code to hijack AI agent transcription processing.
CellPoint secured $34M from Toscafund and launched Zenith, an AI decisioning platform targeting airline and hospitality payments.