Computerworld, Friday, September 18th, 2026
Why AI Companies Are Really Pumping the Brakes on Their Models
Three frontier CEOs called for a slowdown in the same week, but Chinese labs, startups and open-source developers will not follow.
more →
Back Issues/Search Home → Archive → RSS → Subscribe → Popular →
All sections ›
Archive ›
Vol 342 · Issue 3
AI — Archive · Vol 342 Issue 3
27 articles that week
Schneier on Security, Friday, September 18th, 2026
Are AIs Still Struggling With CAPTCHAs?
Anthropic's security-incident transcript shows a frontier model failing a simple image test, possibly mimicking training data.
more →
InformationWeek, Friday, September 18th, 2026
Fall Reading: 10 Expert Takes on AI's Shift From Rollout to Risk Management
A roundup of commentaries from IBM, Walmart and EY leaders on governing agentic systems rather than deploying them.
more →
CIO, Friday, September 18th, 2026
What the McDonald's AI Failure Reveals About AI Agent Mistakes
Three years of drive-thru voice AI failed on accents, dialects and ambient noise, and never beat the cost of a person.
more →
CIO, Friday, September 18th, 2026
Enterprise AI Desperately Needs a Lifecycle for Context
Business definitions drift after launch, leaving AI applications answering with policies legal replaced last week.
more → CIO, Friday, September 18th, 2026
16 Governance Tools for Securing Your AI Fleet
A comparison roundup of guardrail, governance and trust platforms for keeping production LLMs in check.
more → SD Times, Thursday, September 17th, 2026
Why Enterprise Engineering Still Struggles to Prove AI ROI
Token counts and lines of code both failed as proxies for value, pushing teams toward business-unit-level measurement.
more →
InformationWeek, Thursday, September 17th, 2026
AI Panic Is Giving CIOs a New Trust Problem
When frontier labs publicly disagree about AI risk, CIOs should promise process and diligence, never outcomes.
more → CIO, Thursday, September 17th, 2026
AI Agents Should Retrieve Facts, Not Define Them
Connecting an agent straight to the database produced untrustworthy numbers; the fix was removing its authority to define facts.
more → CIO, Thursday, September 17th, 2026
AI-Smart, Not AI-First
AI-first mandates repeat the cloud-first mistake; organizations must be AI-enabled before they can responsibly be AI-first.
more → InfoWorld, Thursday, September 17th, 2026
Your AI Testing Dashboards Are Green. That's the Problem
Self-healing test repair can turn a dashboard into a collection of green opinions rather than evidence.
more →
InfoWorld, Thursday, September 17th, 2026
Stop Tuning Your Models and Fix Your Data
Dropping an LLM into a chaotic warehouse produces a confident idiot; model gains were marginal next to data-foundation work.
more → The Register, Thursday, September 17th, 2026
AI Model Watermarking Changes Agent Behavior
Lasso Security finds SynthID-Text watermarking alters tool calling and safety refusals, especially under prompt injection.
more →
Help Net Security, Thursday, September 17th, 2026
AI Is Adding to the Review Load on Open-Source Projects, Many of Them Thinly Funded
ACM authors warn that AI-generated contributions are outpacing the volunteer maintainers who must vet them.
more →
TechTarget, Wednesday, September 16th, 2026
Guide to AI Data Pipeline Security and Resilience
Organizations must secure AI data pipelines through strategies addressing data poisoning and unauthorized access.
more →
Architecture & Governance Magazine, Wednesday, September 16th, 2026
Enterprise Architecture for Responsible AI: Governance, Risk and Operating Models
Enterprise architecture provides structure for managing AI governance across strategy, data, technology, risk, and operations.
more →
Analytics Insight, Wednesday, September 16th, 2026
Agentic AI Security: Risks, Threats, Best Practices
Autonomous AI agents create new security risks requiring identity controls, tool restrictions, and runtime monitoring.
more →
CSO Online, Wednesday, September 16th, 2026
AI Agent Authorization Risks Remain a Gap in New NIST-CISA Token Security Guidance
NIST IR 8587 covers token forgery and theft but leaves agent actions out of scope, with delegation chains unaddressed.
more →
TechTarget, Tuesday, September 15th, 2026
Why Enterprise AI Fails at the Org Chart, Not the Model Layer
Enterprise AI failures stem from organizational dysfunction and unclear accountability, not technology limitations.
more → InfoWorld, Tuesday, September 15th, 2026
How to Get Better Results From Local LLMs With Ollama
Laptop-class models have improved sharply, and this walks through model choice, parameters and runtime tuning in Ollama.
more → Cybersecurity Dive, Monday, September 14th, 2026
Security Teams Increasingly Outflanked by AI Agents
Non-human AI identities now outnumber human identities by 75:1 in some environments, outpacing security systems.
more →
TechTarget, Monday, September 14th, 2026
AI Harnesses Bring Coordination and Guardrails to Enterprise AI Agents
AI harnesses coordinate multiple agents and provide guardrails for enterprise system interactions.
more → Memeburn, Monday, September 14th, 2026
22 Best AI Tools in 2026: Complete Directory by Category
Comprehensive guide comparing 22 AI tools across seven categories for writing, coding, content creation, and productivity.
more →
Network World, Monday, September 14th, 2026
AI Inferencing Is Headed for the Network Edge
Gartner expects two-thirds of enterprises to deploy edge AI by 2029, up from 10% in 2025, driven by data gravity and sovereignty.
more →
MIT Sloan Management Review, Monday, September 14th, 2026
How AI Creates a Capability Mirage
When polished AI output masks eroding individual skill, teams lose the ability to judge who actually knows what.
more →
InformationWeek, Monday, September 14th, 2026
The Multi-AI Model Stack Is Here. Now Someone Has to Manage It
Put a gateway between applications and models before you scale, and separate standing routing policy from task-level choice.
more → The Observation Deck, Sunday, September 13th, 2026
The Contagion of Fear
Bryan Cantrill argues that AI extinction fears spread irresponsibly despite lacking substantive evidence.
more →