Finance & Payments
Auditors found an AI fraud model's leads outran its reviewers
Australia's Medicare fraud model flagged 597 potential matters and 82 were recorded as reviewed. Its replacement came with a volume dial.
06 October, 2026
AI Agents & Digital Workers
Meta makes a privacy rule pass shadow mode first
Meta lets a privacy rule enforce only after it survives shadow mode, and has published how it would take one back.
05 October, 2026
Finance & Payments
Auditors found AI assessments missing at two tax agencies
The IRS had a rule to stop AI that lacked required assessments, and a team to check. The audit doesn't say if either acted.
05 October, 2026
AI Agents & Digital Workers
Medicare requires clinician review before an AI-assisted refusal
Medicare's AI pilot lets software approve requests alone. A refusal needs a clinician, on the path where vendors can earn.
05 October, 2026
AI Agents & Digital Workers
When the checkpoint stops carrying information
Anthropic planted dangerous commands in test sessions. Reviewers caught 13.6%, fewer the longer they worked.
04 October, 2026
AI Agents & Digital Workers
Cloudflare starts its AI agents with access to nothing
Cloudflare narrows an agent's authority by what it has already read, and an Australian directorate wrote a similar rule.
18 September, 2026
AI Agents & Digital Workers
AI needs the undocumented process written down first
Checking the AI agent's month-end close took as long as doing it, because nobody had written the process down.
10 September, 2026
AI Agents & Digital Workers
Approve, but not merge
Ona, NICE, Brex and Uber gave AI systems authority over a decision. None of them used an autonomy level.
09 September, 2026
AI Agents & Digital Workers
Knowing when authority ends
Approval is only the start. The harder control is proving when one delegated authority has actually finished.
09 September, 2026
AI Agents & Digital Workers
Where the rule lives
An AI agent deleted a production database in nine seconds. The safeguard was real, and attached only to the path a human would take.
08 September, 2026
AI Agents & Digital Workers
A quantity, not a switch
Permission says what an actor may do, never how much. An engineer who could change ten thousand files changes six. An agent does not.
08 September, 2026
Maritime
The Master is ashore
The UK keeps a Master's overriding authority over uncrewed vessels but puts the person ashore. IMO's MASS Code binds nobody until 2032.
04 September, 2026
Industrial Automation
A control you can skip
A welding cell's interlock never failed. It had no opinion about who was inside. A worker died with two supervisors in the cell.
27 August, 2026
Healthcare
Mandated, not measured
Healthcare, and the difference between requiring supervision and verifying it happened. A mandated safety standard nobody measured.
27 August, 2026
Finance & Payments
The one account that stopped the business
SMBs with under 250 employees now account for 63% of all data breaches - more than enterprises by a factor of four. The reason is rarely sophisticated. It is usually one account, one click, and no one whose job it was to notice.
24 August, 2026
Analysis
Fake Compliance as a Service: The Delve Scandal
In March 2026, a whistleblower group exposed a Y Combinator-backed compliance startup allegedly selling fabricated SOC 2 and ISO 27001 reports - with auditor conclusions pre-written before any client submitted evidence. Here is what happened and what it means for organizations relying on compliance automation.
17 August, 2026
Energy & Utilities
When Feedback Misses Its Window
The Iberian blackout of April 2025 had a closed-loop system with feedback, protection, and reactive power reserves. The correction mechanism existed and sat unused while voltage climbed toward collapse. The loop existed. It didn't govern.
22 July, 2026
Autonomous Vehicles
Supervision Can't Be Bolted On
Most people think an autonomous system is safe as long as someone can always intervene. The Cruise robotaxi incident shows why that assumption fails - and what proactive supervision actually looks like when reactive supervision isn't enough.
21 July, 2026
AI Agents & Digital Workers
What a Real AI Supervisor Actually Requires
AI agents move in discrete steps, and the question isn't how fast you catch a problem - it's how many unsupervised steps an agent takes before anyone is in a position to catch anything. Control engineering solved this in 1987.
15 July, 2026
AI Agents & Digital Workers
Managed Autonomy, Not Autonomous AI
Companies solved the AI agent governance problem for humans decades ago - it's called Privileged Access Management. Most AI agents were never enrolled in it, and where the analogy breaks is the actually interesting part.
15 July, 2026
AI Agents & Digital Workers
Semi-Autonomous AI
When AI agents act without being asked, the question stops being whether they made a good decision. It becomes whether anyone can even say which piece of software made it - and what you can actually do about it.
13 July, 2026
AI Agents & Digital Workers
Open-Loop, Closed-Loop, and the Line Between Hoping and Governing
A system that fires an instruction and hopes is not the same as one that watches itself and corrects. Most of what's being called AI governance right now is the first one, wearing the second one's badge.
13 July, 2026
AI Agents & Digital Workers
The Systems We Trust Are Never Fully Autonomous
Autonomy plus control isn't a compromise. It's the answer to AI anxiety. Every technology humans have ever trusted with something that mattered got trusted because a control layer was built alongside it.
06 July, 2026