VentureBeat

85% of enterprises run AI agent pilots. Only 5% reach production. Cisco's Michael Dickman says the gap is identity governance — and it has to be solved at machine speed. venturebeat.com/ai/cisco-dickm…

AI tool poisoning exposes a major flaw in enterprise agent security venturebeat.com/ai/ai-tool-poi…

Intent-based chaos testing is designed for when AI behaves confidently — and wrongly venturebeat.com/ai/intent-base…

Fully driverless. No safety driver. Real city. That was Rashed Haq. Now he's VP of Autonomous Vehicles at @GM and speaking at #VBTransform 2026, the event for enterprise AI leaders who are done piloting. 📅 July 14–15 → bit.ly/49oTA6r

Anthropic says it hit a $30 billion revenue run rate after 'crazy' 80x growth venturebeat.com/ai/anthropic-s…

Context ceilings made voice agents expensive to orchestrate. OpenAI's three new Realtime models are built to fix that. venturebeat.com/ai/openai-brin…

RedAccess scanned vibe coding platforms and found 380,000 exposed corporate apps. 5,000 held sensitive data. Your SIEM never saw any of them. venturebeat.com/ai/vibe-coded-…

Pedro Franceschi has been breaking rules and building things since he was 13. First iPhone jailbreak in Brazil → $1.5B fintech at 15 → co-founder of Brex, now behind 35,000+ companies worldwide. He's speaking at VB Transform 2026. See you there. 📅 July 14–15 | Hotel Nia, Menlo Park → bit.ly/49oTA6r

A Fortune 50 AI agent rewrote a security policy with valid credentials and authorized access. Cisco's 6-stage identity model is the fix. venturebeat.com/ai/cisco-crowd…

Anthropic wants to own your agent's memory, evals, and orchestration — and that should make enterprises nervous venturebeat.com/ai/anthropic-w…

Governance, not gatekeeping: How SAP brings enterprise‑grade safety to AI connectivity venturebeat.com/ai/governance-…

Enterprises spent $401 billion on AI infrastructure. Average GPU utilization: 5%. The bill is due and the CFO is paying attention. venturebeat.com/ai/5-gpu-utili…

Anthropic introduces "dreaming," a system that lets AI agents learn from their own mistakes venturebeat.com/ai/anthropic-i…

Sakana AI trained a small model to orchestrate GPT-5, Claude Sonnet 4, and Gemini 2.5 Pro dynamically — no hardcoded pipelines, 6x fewer tokens than competing frameworks in the researchers' tests. venturebeat.com/ai/how-sakana-…

Meet ZAYA1-8B, a super efficient open reasoning model trained on AMD Instinct MI300 GPUs venturebeat.com/ai/meet-zaya1-…

Why AI breaks without context — and how to fix it venturebeat.com/ai/why-ai-brea…

Anthropic Skill scanners passed every check. The malicious code rode in on a test file. Here's the seven-surface audit grid and three CI fixes to deploy now. venturebeat.com/ai/anthropic-s…

SaaS companies lost up to 80% of their market value in months. Jerry Liu on what's left when agents eat the interface. youtube.com/watch?v=HbXvX-…

Speaker #2 for AI Impact Series NYC (May 20) 👇 Andrew Reiskind, Chief Data Officer @Mastercard Data strategy + AI + regulation at global scale. He'll be in the room with Preeti Rathi of @temporalio for a real conversation, not a keynote. Small room. Limited seats. NYC. 👉 bit.ly/4u13C6g

Market research is too slow for the AI era, so Brox built 60,000 identical 'digital twins' of real people you can survey instantly, repeatedly venturebeat.com/ai/market-rese…