AI is moving faster than governance can keep up, and the cracks are starting to show. This edition we look at why agentic AI adoption is outpacing the frameworks needed to control it, how data quality has become the foundation every AI system depends on, and what a leaked Anthropic model tells us about where the frontier is heading next.
The trend: Most enterprises are still in the pilot phase with AI, but a rapid shift to production-scale agentic deployment is now underway - and governance is struggling to keep pace.
The details: Deloitte's 2026 State of AI in the Enterprise report found that access to AI has grown 50% in a single year, but fewer than 60% of workers actually use it daily. Only 34% of companies are deeply reimagining their business around AI, while 37% are still using it at a surface level with no real process change. The agentic AI picture is striking: only 23% of companies are using it at even a moderate level today, but 74% expect to be doing so within two years. Yet just 21% currently have a mature governance model for autonomous agents.
Why it matters: Agentic AI doesn't wait for human review — it acts. That means the gap between adoption and governance readiness isn't just a compliance concern; it's an operational risk that compounds with every agent you deploy. Clean, governed, real-time data pipelines are the only thing standing between an autonomous system and an autonomous mistake.
The trend: Data quality has moved from a back-office concern to a frontline AI priority, and the market for platforms that support it is being rebuilt around AI-native capabilities.
The details: Forrester's Q1 2026 Data Quality Wave report found the market shifting away from rule-based cleaning toward platforms that embed generative and agentic AI. Four trends are reshaping buyer expectations: AI integration across the data quality lifecycle, real-time observability replacing static validation, multimodal support for unstructured data like documents and images, and unified platforms replacing fragmented point solutions.
Why it matters: As organizations scale generative and agentic AI, data quality sits at the foundation of whether those systems can be trusted. Poor data quality doesn't just produce bad reports, it produces bad decisions at scale.
A new Pew Research Center survey found that a majority of Americans view data centers negatively; 39% say they are mostly bad for the environment, 38% say they harm home energy costs, and 30% say they hurt quality of life for people living nearby. Crucially, those most familiar with data centers hold the most negative views, and younger adults are more critical than older ones. The backlash has reached Capitol Hill, with Senator Bernie Sanders introducing legislation to halt all new data center construction in the US, arguing that Congress has no framework to protect the public from the speed of AI infrastructure expansion.
Details of Anthropic's most powerful model to date surfaced this week after a CMS configuration error left thousands of unpublished assets, including a draft launch blog, in a publicly searchable data store. The leaked post describes a new "Capybara" model tier, sitting above Opus, and calls Mythos "by far the most powerful AI model we've ever developed," with particular concern flagged around its cybersecurity capabilities, which Anthropic says are "currently far ahead of any other AI model." The company confirmed to Fortune that a new general-purpose model representing "a step change" is currently in early access testing.
%201.svg)
