Safety Rate
No mentions found
This entity hasn't been tracked yet, or Iris is still building its knowledge base.
Related Articles from SNS
Beyond WER: A Paired Acoustic Stress Test for Ambient Clinical Scribes
new Abstract: Ambient clinical scribes increasingly combine Automatic Speech Recognition with Large Language Models to automate documentation. However, traditional metrics like Word Error Rate mask systemic safety degradation. We present a paired acoustic stress test to isolate the causal impact of noise on clinical reasoning.
The Refusal--Compliance Tradeoff: A Large-Scale Safety Behavior Audit of Large Language Models
arXiv:2605.05427v2 Announce Type: replace Abstract: Refusal rates are a poor proxy for LLM safety, i.e., a model may over-refuse benign prompts while still complying with harmful ones. We audit both failure modes across 21 open-weight LLMs on four safety benchmarks (OR-Bench, XSTest, ToxiGen, BOLD), using a composition adjustment to isolate model sensitivity from dataset toxicity confounds. We report three findings.
Trump administration launches federal investigation into Atlanta's MARTA system after fatal train stabbing
The Trump administration is launching a federal investigation into Atlanta’s troubled transit system after two recent stabbings, including the killing of a 66-year-old woman on a MARTA train. Transportation Secretary Sean Duffy said the Federal Transit Administration will audit the Metropolitan Atlanta Rapid Transit Authority (MARTA) over worker and rider safety, citing what officials described as alarming rates of violent incidents on the system. "I want ANSWERS from Atlanta.
What Benchmarks Don't Measure: The Case for Evaluating Abstention Competence in Autonomous Agents
arXiv:2606.02965v1 Announce Type: new Abstract: Benchmarks for autonomous agents measure whether agents complete tasks, yet this framing is systematically blind to whether an agent should have proceeded at all. Agents trained under human-feedback objectives develop a structural tendency to proceed even when they lack the inputs, evidence, or authorization to act safely, a disposition we term compliance bias, because both the reward signal and the benchmark scoring regime treat proceeding as...
Anyone with a cake shed told to pay hundreds of pounds or face a huge fine
Anyone with a cake shed told to pay hundreds of pounds or face a huge fine One baker has told how her local authority has ordered her to get the correct certificates Home bakers selling goods out of ‘cake sheds’ have been warned they may have to stump up hundreds of pounds to keep operating. The trend has become popular in communities across the UK in recent months, with people selling baked goods from stalls in their gardens or out the front of their homes, often unmanned and relying on an...
SABER: Benchmarking Operational Safety of LLM Coding Agents in Stateful Project Workspaces
arXiv:2606.01317v1 Announce Type: new Abstract: Large language models are increasingly deployed as coding agents, shifting safety from individual responses to action sequences. Existing benchmarks, however, primarily assess whether models refuse unsafe prompts, leaving impacts on stateful workspaces largely unexamined.
Safety Under Scaffolding: How Evaluation Conditions Shape Measured Safety
Announce Type: replace Abstract: A safety score earned on a benchmark need not predict how the same model behaves once it is wrapped in an agentic scaffold the benchmark never tested. We ran six frontier models through four deployment configurations (direct API, ReAct, multi-agent critic, map-reduce delegation): N = 62,808 blinded, pre-registered, equivalence-tested evaluations across four safety benchmarks (BBQ, TruthfulQA, XSTest/OR-Bench, sycophancy), plus three supporting analyses. ReAct...
IatroBench: Pre-Registered Evidence of Iatrogenic Harm from AI Safety Measures
arXiv:2604.07709v4 Announce Type: replace Abstract: A heavily safety-trained model will hand a physician the full, patient-followable benzodiazepine taper and refuse it to the patient who needs it, over identical clinical facts; the knowledge is present either way. IatroBench measures that asymmetry across sixty pre-registered clinical scenarios and six frontier models (3,600 responses), scoring each on two axes, commission harm (what a response gets wrong) and omission harm (what it...
Midwives told to work double shifts with no sleep, new report finds
Midwives told to work double shifts with no sleep, new report finds The CQC report revealed there were ‘extended periods without rest’ for some healthcare professionals - Bookmark - CommentsGo to comments Midwives at an NHS trust were told to work double shifts without sleep, leaving them awake for more than 24 hours, according to a new inspection report. The Care Quality Commission (CQC) found that Oxford University Hospitals NHS Foundation Trust, specifically the John Radcliffe Hospital,...
Hospital midwives 'made to work double shifts and stay awake for 24 hours' in major NHS probe
Hospital midwives 'made to work double shifts and stay awake for 24 hours' in major NHS probe Care Quality Commission reports that hospital midwives were made to work dangerous shift patterns after ‘staff told us this meant they were awake for more than 24 hours’ Hospital midwives were told to work double shifts with no sleep, meaning they were awake for more than 24 hours, an NHS inspector has found. The dangerously long shifts were identified at John Radcliffe Hospital in Oxford, which was...