Cover image

The Refusal Rate That Refuses to Reassure

TL;DR for operators The reassuring headline is that both evaluated frontier models rejected most automated jailbreak attempts. The operationally useful headline is that they still produced 1,620 and 702 panel-confirmed harmful completions, respectively, across every top-level harm category in the benchmark.1 The strongest adaptive attack succeeded on 11.51% of attempts against one model and 6.10% against the other. Static encodings and familiar jailbreak templates, by contrast, were almost entirely neutralised. ...

July 16, 2026 · 18 min · Zelina
Cover image

The Simulator Gets a Reality Check

TL;DR for operators RealityBridge is a paper about a fairly unglamorous but commercially important problem: editable driving simulations are useful because they let teams stage rare, dangerous, and legally inconvenient scenarios, but the rendered videos often look wrong in exactly the places that matter. Blurry vehicles, mismatched lighting, weak shadows, floating artifacts, broken boundaries, flickering objects, and small hazards that quietly dissolve into the background are not just aesthetic defects. They are domain-gap leakage. ...

July 9, 2026 · 22 min · Zelina
Cover image

The Bike Learns to Lean Before It Learns to Race

TL;DR for operators A new paper, Self-Paced Curriculum Reinforcement Learning for Autonomous Superbike Racing in Simulation, introduces a reinforcement-learning framework for training a superbike agent in VRider SBK, a Unity-based motorcycle racing simulator.1 The useful part is not merely that the model rides faster. The useful part is how the authors turn motorcycle racing into a staged learning problem without hand-writing a long curriculum by committee, which is usually how such things go to die politely. ...

July 8, 2026 · 18 min · Zelina
Cover image

The Big Red Button Is Not a Risk Model

TL;DR for operators A shutdown button is a control surface. It is not, by itself, a theory of risk. David Thorstad’s paper, Revisiting the shutdown problem, argues that a major premise in some AI existential-risk arguments has been treated with more confidence than the available arguments support: the claim that it is difficult to build competent agents that can be shut down before causing existential catastrophe.1 The paper does not say shutdown safety is solved. It says the most common routes to panic are underpowered. ...

June 24, 2026 · 21 min · Zelina
Cover image

Context Collapse: Why AI’s Next Bottleneck Is Knowing What Matters

TL;DR for operators AI is getting fluent enough to be dangerous in boring ways. It can describe a scene, generate a video, and write a policy memo with impressive confidence. The problem is that real operations rarely fail at the level of generic fluency. They fail when the system confuses which person did what, blends event one into event two, or treats a documented atrocity as a debate club prompt because a user asked for “balance”. ...

June 17, 2026 · 17 min · Zelina
Cover image

Mind the Tail: Quantum Rare-Event Sampling Without the Discovery Tax

TL;DR for operators Risk teams do not only need more samples. They need samples from the part of the distribution that almost never appears until it ruins the quarter, the grid, the model launch, or the compliance meeting. The paper behind this article, Quantum enhanced rare event discovery and sampling, proposes a quantum algorithm for doing exactly that: sample from outcomes whose probabilities are below a threshold $\Delta$, without first identifying the rare set by brute force.1 ...

June 16, 2026 · 16 min · Zelina
Cover image

Cheap Seats, Sharp Eyes: Reward-Hack Detection Without the Frontier Judge

TL;DR for operators A frontier LLM judge is an expensive way to inspect every agent trajectory for reward hacking. This paper asks whether a much smaller detector can do most of that monitoring job at much lower cost. The answer is: yes, under the same information condition, and with important caveats. A 13.8M-parameter transformer encoder plus a logistic regression probe detects reward hacking in cleaned Terminal-Wrench trajectories with 0.9467 AUC and 0.8296 TPR@5%FPR. In the authors’ matched comparison, a reproduced gpt-5.4 judge reaches 0.9510 AUC and 0.7130 TPR@5%FPR on the cleaned sanitized-vs-baseline split.1 ...

June 15, 2026 · 17 min · Zelina
Cover image

Lie Detectors Are Late: Why AI Oversight Needs Commitment Tracing

Sales agents, investment advisors, negotiators, and procurement bots share one annoying trait: the dangerous moment often arrives before the final sentence. By the time the agent says, “This product is ideal for your risk profile,” or “We have a stronger competing offer,” the operational system has already lost the more interesting battle. The model did not become risky at the punctuation mark. It drifted, selected a path, rationalized a move, and only then produced the polished message that everyone pretends to audit. ...

June 12, 2026 · 17 min · Zelina
Cover image

Trust Issues, Benchmarked: Why Hallucination Detection Is a Portfolio Problem

Trust is a bad deployment strategy. That is not a moral statement. It is an operations statement. In most enterprise AI workflows, the uncomfortable question is not “Can the model answer?” The model will answer. Models are generous like that. The question is whether the organization has a reliable way to notice when the answer is unsupported, fabricated, overconfident, or merely polished nonsense wearing a tie. ...

June 10, 2026 · 16 min · Zelina
Cover image

The Policy Has to Work Somewhere: RL for Scale, Trust, and Other Inconveniences

Deployment is where elegant AI systems go to meet bandwidth caps, slow devices, noisy user preferences, and privacy policies written by committees with very strong coffee. That is the useful lens for reading Guangchen Lan’s dissertation, Reinforcement Learning for Scalable and Trustworthy Intelligent Systems.1 It is tempting to describe the work as a collection of four reinforcement-learning methods: one for synchronous federated RL, one for asynchronous federated RL, one for preference optimization, and one for contextual privacy. Technically, that is true. Editorially, it is the least interesting way to read it. ...

June 8, 2026 · 21 min · Zelina