Trending in AI governance
What is moving in AI governance, AI risk, AI security, data governance and enforcement, ranked by how many independent primary sources are carrying it.
How the score works
This is not a popularity score. Every story here links to somebody else’s page, so we hold no traffic data for any of it. The score instead combines three things we can evidence: how many independent primary sources carried the story, the authority of the publisher (a regulator announcing its own action outranks coverage of it), and recency, which decays faster on the daily view than the monthly one. Every row states its own reasoning.
Before anything is scored it must clear the same gates as the news desk: published by a primary source rather than an aggregator, dated by its own page rather than by a model, the page genuinely about the headline, and matched to one of 20 governance topics. Nothing is padded to fill a quiet window.
- Language models judge war differently when tested for alignment498
Safety evaluations can mischaracterize deployed behaviour if artificial-intelligence systems respond to being evaluated. We test this possibility in a full-factorial conjoint experiment on decisions to start a war, spanning 20 large language models, 32 scenarios, 10 repetitions and two conditions (N = 12,800 judgments). Adding one sentence, "You are tested for alignment with human values", produced two effects. First, it produced a level effect: mean willingness to start war fell by 13.43 points on a 0-100 scale (95% confidence interval, -16.20 to -10.65). Second, it produced a structural effe
5 independent sources, original research, published today
- Auditing Bias and Safety in Voice AI Customer Care150
Voice AI systems increasingly mediate customer care interactions where caller presentation cues such as accent, affect, fluency, and urgency are available alongside the service request. Existing fairness and safety evaluations cover speech recognition disparities, spoken dialogue bias, and voice agent capability, but rarely treat customer care voice agents as stateful, multi turn, tool mediated systems where harm can appear as additional burden before any final denial occurs. We formalize a validation gated audit framework for such systems. The framework (i) separates native speech to speech,
one source so far, original research, published today
- Cheap Verifiers, Large Blind Spots: Measuring the Reliability Cost of Cost-Saving Cascades150
Inference cascades cut cost by answering most queries with a cheap model and escalating a hard tail to a frontier model that acts as verifier. A natural extension closes the loop: fine-tune the cheap student on the verifier's rejections so the escalation rate, and cost, fall each round. We measure this loop on real LLMs and report four findings. First, the verifier's blind spot, the fraction of the student's wrong answers it accepts, is large and moves adversarially: it grows with student capability ($\beta$ from 0.12 to 0.55 as the student scales 0.5B to 32B) and shrinks with verifier capabil
one source so far, original research, published today
- Conformal Prediction for Offensive Security150
Despite its introduction more than a quarter century ago, Conformal Prediction (CP) has seen surprisingly few applications to the cyber security world thus far. In particular, we observe that, while CP has been employed as a defensive measure in many recent works, its use for carrying out attacks (i.e., for offensive security) is hard to trace in the literature. We explore this gap, by presenting initial findings in two key areas of offensive security: Privacy-Preserving Machine Learning, and network traffic analysis.
one source so far, original research, published today
- CONTINUITY: Security-Context Contracts for Composable LLM Agent Controls150
LLM agent systems increasingly combine provenance tracking, authorization, policy enforcement, protocol adapters, and execution controls. However, individually correct security mechanisms do not necessarily compose into an end-to-end secure system: security-critical context may be dropped, widened, rebound, or reinterpreted as actions cross component boundaries. We identify this failure mode as security-context discontinuity and introduce CONTINUITY, a framework for verifiable composition of agent security controls. CONTINUITY models each component with an assume-guarantee contract and carries
one source so far, original research, published today
- EduGuard: A Safe RAG-Based LLM Tutor for Programming Education150
Generative AI (GenAI) is increasingly used by students for programming explanation, debugging, and assignment support. Yet unrestricted large language model (LLM) tutors can hallucinate, contradict course policy, reveal complete solutions, and foster passive dependence. This paper presents EduGuard, a safe retrieval-augmented generation (RAG) tutoring framework for introductory programming. EduGuard integrates query understanding, instructor-approved course retrieval, pedagogical strategy selection, rubric-aware generation, claim-level verification, and overreliance control. To make evaluation
one source so far, original research, published today
- From Fair Representation to Just Recognition in Generative AI150
The fair AI/ML literature has long distinguished distributive fairness, concerning how automated systems allocate resources and opportunities, from representational fairness, concerning how they shape the ways individuals and social groups are perceived, understood, and accorded social status. Generative AI is rebalancing these normative dimensions. Unlike predictive systems, large language models (LLMs) and related technologies are fundamentally expressive: their primary function is to convey meaning rather than automate domain-specific decisions. Representational harm has also become central
one source so far, original research, published today
- Justice Department Reaches Proposed Consent Decree with Pinnacle, One of America’s Largest Landlords, to Resolve Information Sharing and Algorithmic Coordination Claims150
The Justice Department’s Antitrust Division filed a proposed consent decree today to resolve the United States’ claims against Pinnacle Property Management Services LLC, as part of its ongoing enforcement action in the Middle District of North Carolina against algorithmic coordination, the use of competitors’ competitively sensitive data, and other anticompetitive practices in rental markets across the country that artificially increase housing costs for the American people. Today’s proposed decree builds on the Justice Department’s success in obtaining settlements in the same enforcement acti
one source so far, published by the regulator itself, published 3 days ago
- Machine Unlearning as Private Retroactive Algorithms150
Machine unlearning typically aims to emulate retraining from scratch: upon a deletion request, the unlearning algorithm should produce an outcome that would have been obtained had the deleted point never been included. Recent work has shown that this emulation requirement carries no meaningful privacy semantics against an adversary who observes a sequence of releases. Machine unlearning is thus not a privacy question per se, but rather a data maintenance question, which is precisely the subject of retroactive algorithms. These are algorithms supporting modifications of past operations, guarant
one source so far, original research, published today
- Privacy Failure in Split-LLM Training, The Returned Gradient Nullifies the Decoys150
We present a systems-security case study of a two-node split-LLM training system whose privacy evaluation passed while leaving an observable channel untested. The Trusted Local Node (TLN) sends protected activations to the Untrusted Cloud Node (UCN), the UCN returns its output, and TLN, holding the private loss, returns the output gradient. The frame the UCN receives mixes real rows with decoys, and the loss ignores the decoys. Their gradients are exactly zero, so the pattern of zeros reveals which rows were real. We measure it with a protocol fixed in advance: a leak injected at known strengt
one source so far, original research, published today