Latest posts
-
Physical AI: Inner Sky Labs’ Safety-First Robot Stack
Inner Sky Labs, the maker of the Miko robot, has launched a full physical AI stack with an independent safety layer, Kavach — and is closing a ₹300 crore round. We look at what it proves, and the gap between language safety and physical safety.
-
Content provenance: can AI be trained on licensed culture?
Eros Innovation and Innovate UK Business Connect are funding content provenance tools for cultural AI. The question is real and now regulated by the EU AI Act. The sponsor’s headline claim — a $3bn “independently valued” dataset — is weaker than it sounds.
-
Enterprise AI Governance: authority moved, accountability didn’t
Thoughtworks surveyed 3,200 CIOs and found authority over AI distributed across the business while accountability stayed with central IT. Independent research from EY, OneTrust, ABBYY and Grant Thornton tells the same story.
-
AI Agent Governance: the 40% of agents enterprises will switch off, and the 21% that are ready
Adoption of autonomous AI agents has outrun the controls meant to govern them. This analysis maps the widening gap between fast deployment and weak oversight — through Gartner’s 40% decommission forecast, a CISO visibility crisis, documented incidents at ServiceNow and Air Canada, and the EU and NIST timelines now in force.
-
AI Assurance Goes to Market: Apexon and TrustModel.ai Bet on the Trust Layer for Enterprise AI
Apexon and TrustModel.ai are selling AI assurance as the trust layer for enterprise AI — 0-100 TrustScores, behavioural agent certificates, and supplier AI risk assessed without consent. The category is real; the authority is not yet earned.
-
Proofpoint Agentic Security: Two Launches, One Knowledge Graph, and Questions Left Open
Two Protect 2026 launches, one Knowledge Graph: Proofpoint wraps 18 months of acquisitions into an agentic security story — while rivals ship comparable pieces sooner.
-
AI Agent Governance Meets Product Delivery: Info-Tech Warns Without Showing Its Data
A research firm warns that AI agents now influence product decisions without human authorization — and cites no evidence of its own. The data exists; none of it is in the release.
-
Generative AI 2026: We Fact-Checked the Definitive 360° Survey. Here’s What Held Up.
A 54-citation survey of generative AI gets adoption, regulation and greenwashing right — but its frontier models are 18 months stale and its environmental numbers mix AI with all data centres.
-
Cloudflare AI Crawler Controls Go Live: Search Stays, Training Can Be Refused
Cloudflare’s deadline to the AI industry arrived: keep search, refuse training. But the controls bind only 8.5% of the top web, and the referee sells the tollbooth.
-
Agentic AI vs Generative AI: The Budget Mistake Draining Enterprise AI Spend
New research argues that treating agentic AI and generative AI as the same thing quietly drains budgets. The headline failure statistics check out—but the research misses the documented successes that show exactly what the surviving 60% do differently.