Search: “policy-debate”

Smart search ranks by meaning as well as keywords (one row per story, last 45 days).

MInTRL: Off-policy Intervention can boost On-policy RL

MInTRL injects sparse judge corrections into on-policy RL rollouts, expanding exploration beyond on-policy sampling while preserving learnability on math and code benchmarks.

Minimal Intervention Reinforcement Learning periodically has a judge-intervention policy replace erroneous suffixes of the current policy's output with short corrections, then returns control, keeping trajectories largely on-policy. Training uses a sequence-level advantage-regression objective that removes the need for importance sampling. Across math and code benchmarks it consistently beats standard on-policy and off-policy baselines, remains effective with self-intervention, and performs best at moderate intervention intensity.

Hugging Face daily papers·Sep 10, 20:00 UTC · 6d agoAI research

40

The Guardrails Debate: Security Researcher Changes His Mind

A security researcher revises his stance on AI guardrails, arguing defenders need more help as attackers ignore safety rules.

Dark Reading publishes an opinion piece in which a security researcher changes his mind about AI guardrails. He acknowledges guardrails are critical, citing recent high-profile incidents, but argues defenders need more support because attackers do not play by the same rules. The piece contributes to the ongoing debate on defensive AI controls rather than reporting a new incident.

Dark Reading·Aug 31, 20:09 UTC · 16d agoAI safety & security

24

One Symptom, Three Levers: A Critical Review of On-Policy Self-Distillation

A review paper frames on-policy self-distillation collapse as governed by three levers: token weighting, privileged information, and guidance decay.

The paper critically reviews On-Policy Self-Distillation (OPSD), where a language model trains on its own generations scored token-by-token by a teacher conditioned on privileged information such as reference solutions or environment feedback. It identifies collapse, the progressive narrowing of producible reasoning paths, as the dominant failure mode and analyzes it through three levers: signal weighting, the nature of privileged information, and teacher dynamics. The review is restricted to mathematical reasoning, reports no new experiments, and offers a shared vocabulary separating settled findings from disputed ones.

Hugging Face daily papers·Aug 25, 20:00 UTC · 22d agoAI research

45

Before You Poll with LLMs: A Deliberative Diagnostic Framework

Deliberative diagnostic shows all five tested frontier LLMs misrepresent human belief shifts after arguments, with GPT-5.1 reversing on outgroup questions.

The Deliberative Polling Diagnostic Framework compares human and LLM persona belief shifts after identical informational interventions, using data from America in One Room (526 personas, 72 questions). All five frontier models tested failed uniquely: GPT-5.1 exhibited partisan reversal (80% on outgroup vs 26% on policy questions), Gemini 2.0 Flash, Claude Sonnet 4.5 and Llama 3.3 70B overshot at 5-7x human magnitude, and DeepSeek V3 showed near-zero change (rigidity). The authors term the underlying signature 'self-sycophancy', conformity to the model's internal persona stereotype rather than reasoning from provided information.

arXiv cs.AI / cs.LG / cs.CL·Sep 14, 16:41 UTC · 2d agoAI research

40

Eliciting Weak-to-Strong Generalization with On-Policy Reverse Distillation

OPRD distillation enables weak-to-strong generalization by amplifying verifier-supported policy updates, outperforming existing RL and distillation methods with fewer student updates.

On-Policy Reverse Distillation (OPRD) evaluates a weak teacher's policy shift relative to its reference policy on student rollouts and amplifies the verifier-supported component of the student's policy gradient. This rescaling preserves the stationary points of policy optimization while letting the student learn beyond the teacher's capacity ceiling. In successive model transfer and multi-teacher distillation, OPRD achieves higher performance with fewer student updates than existing RL and distillation approaches, and response-style analysis shows students remain closer to verifier-RL-trained models than to their weak teachers.

Hugging Face daily papers·Sep 7, 20:00 UTC · 9d agoAI research

33

Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization

ModerationBench shows foundation models can nearly triple Bluesky's moderation F1 (0.60 vs 0.22), with instruction- and example-driven guidance performing comparably.

Researchers built ModerationBench, a new benchmark of 4,000 manually annotated in-the-wild posts from Bluesky, to test whether foundation models can reliably operationalize content moderation policies. They systematically compare instruction-driven guidance (reasoning from policy precepts) with example-driven guidance (generalizing from precedents) for Vision-Language Models. Both paradigms achieve comparable peak effectiveness, and foundation models nearly triple the F1 of Bluesky's deployed moderation system on Random Posts (0.60 vs 0.22).

arXiv cs.AI / cs.LG / cs.CL·Sep 9, 16:30 UTC · 7d agoAI research1

28

The AI policy window is open. We need to act.

OpenAI calls for mandatory national AI safety regulation and backs four California AI safety bills as capabilities accelerate.

OpenAI argues the rapid pace of AI progress, including signs of AI-accelerated research, requires urgent policy action through mandatory, capability-based national regulation. The company endorses four California bills (SB 813, AB 1405, SB 1119, AB 1864) covering independent safety assessments, AI auditor standards, youth protections, and safeguards against AI-enabled biological threats. It also commits to industry-led frontier standards, international coordination, and strengthening internal safeguards such as universal trajectory monitoring and mandatory alignment-evaluation gates for its Astra model. The post references chief scientist Jakub Pachocki's warning about recursive self-improvement and Greg Brockman's "defenders window" concept.

OpenAI News·Sep 9, 13:00 UTC · 7d agoAI policy

72

The Race to Control AI and Protect What Makes Us Human

Opinion piece surveys the AI existential-risk debate, citing Bill Gates' memo and Anthropic's Evan Hubinger on unsolved superintelligence alignment.

A SecurityWeek opinion piece debates whether AI will be a force for good, anchored on Bill Gates' 6,000-word August 2026 memo warning of a turbulent, under-prepared AI transition. Anthropic alignment lead Evan Hubinger stated he believes there is a greater than 10% chance AI kills all humans within a decade and that no plan exists to solve superintelligence alignment. The piece also notes OpenAI reportedly slowed parts of model development over safety concerns and Gates' warning that heavy AI use is associated with reduced critical thinking.

SecurityWeek·Sep 14, 13:00 UTC · 2d agoAI safety & security

25

Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic

Multiverse Computing's Hugging Face post argues language models should refuse only the relevant subset of a topic instead of over-refusing whole subjects.

A Hugging Face blog post by Multiverse Computing examines refusal granularity in language models, arguing models should refuse the relevant subset of a topic rather than the entire topic. No full article text was available for additional technical detail.

Hugging Face Blog·Sep 8, 14:23 UTC · 8d agoAI safety & security

20

Generalized Agent Iteration: One Formal Framework for Iterative Policy Improvement and Recursive Self-Improvement

Generalized Agent Iteration formally unifies iterative policy improvement and recursive self-improvement, defining axes that distinguish anchored, goal-drifting, and self-referential agents.

The paper proposes Generalized Agent Iteration (GAI), a formal framework that models learning as a cycle of agent evaluation and agent improvement, defining the agent as a configuration of modifiable components. Two dials—whether the improving mechanism is part of the agent and whether the evaluation standard is grounded outside it—separate generalized policy iteration (GPI) from recursive self-improvement (RSI) and classify systems as anchored, goal drift, or fully self-referential. The framework places existing systems on shared axes and makes defects of recursive self-improvement statable one condition at a time.

Hugging Face daily papers·Sep 10, 20:00 UTC · 6d agoAI research

30

ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation

ActReview post-trains Qwen3-8B-Base on OpenReview rebuttals to generate actionable peer-review feedback with grounded revision suggestions, benchmarked on 1,000 curated instances.

The paper defines Actionable Peer-review Generation as diagnostic claim generation plus revision suggestion generation and introduces ActReview, a rebuttal-guided post-training framework. From OpenReview review-rebuttal threads the authors build ActReview-40K, aligning reviewer weaknesses with author responses grounded in localized paper evidence, and post-train Qwen3-8B-Base with multi-task SFT followed by GRPO using weakness-specific rubric rewards. They also release ActReview-Bench, a human-curated 1,000-instance benchmark, on which ActReview outperforms prior specialized review-generation models on actionability and grounding while remaining competitive with strong prompt-based LLMs. Human evaluation confirms improved revision usefulness but identifies a remaining gap in technical accuracy.

Hugging Face daily papers·Sep 7, 20:00 UTC · 9d agoAI research

30

Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails

Research shows on-policy expert correction, not imitation fine-tuning, lets weaker agent models catch up under evolved harnesses.

Researchers study how to combine automated agent-harness evolution with lightweight fine-tuning across seven enterprise agent tasks. Naively training weaker models (Qwen3-Coder, Gemma 4) on expert trajectories under an evolved harness regressed performance by 4 to 30 points on all tasks. They propose an on-policy correction pipeline, automated by a meta-level MLE agent, where an expert rewrites only the failing turn of the weaker model's rollout, preserving model-harness fit.

arXiv cs.AI / cs.LG / cs.CL·Sep 8, 17:53 UTC · 8d agoAI research

35

MasterControl Seventeen Every Time

Governed enterprise analytics study shows deterministic policy execution matched 110/110 answer-and-evidence contracts while runtime agent planning matched none.

The paper studies a governed approach where a language model interprets the question while deterministic policy selects and runs a pre-approved analytical program returning results and evidence. Across 440 runs, three 8B models generated SQL and selected tools at runtime, while Qwen3-8B only interpreted intent and policy executed the approved program. None of 330 runtime-planning episodes satisfied the full answer-and-evidence contract, whereas the policy-executed analyzer matched 110 of 110. The authors note this is configuration-specific and expressiveness is preserved via relational operations, aggregation, comparison, windows, ranking, and similarity with replayable results.

Hugging Face daily papers·Sep 1, 20:00 UTC · 15d agoAI research

24

ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation

ActReview post-trains Qwen3-8B-Base on 40K rebuttal-derived instances with rubric rewards to generate actionable, grounded peer-review feedback, plus a 1,000-instance benchmark.

The framework builds ActReview-40K from real OpenReview review-rebuttal threads, aligning reviewer weaknesses with author responses and grounding feedback in localized paper evidence. Qwen3-8B-Base is post-trained with multi-task supervised fine-tuning followed by GRPO using candidate-aware, weakness-specific rubric rewards. Experiments show improved actionability and grounding over prior specialized review-generation models, supported by ActReview-Bench, a human-curated 1,000-instance evaluation set. Human evaluation confirms better revision usefulness while noting a remaining gap in technical accuracy.

arXiv cs.AI / cs.LG / cs.CL·Sep 8, 17:30 UTC · 8d agoAI research

31

Agent as Policy for Robotic Manipulation

Agent as Policy lets a general-purpose agent drive a physical robot via runtime reasoning and program generation, reaching 100% success on manipulation tasks.

The paper introduces Agent as Policy (AGP), which puts task planning and execution for a physical robot under a general-purpose agent's control with no task-specific or environment-specific training. The agent interprets visual evidence, writes executable programs, issues motion commands, and revises actions based on physical outcomes. AGP was evaluated on real-world manipulation tasks including assembly from human videos, block construction from goal images, die reorientation, targeted throwing, and bimanual towel folding. It achieved success rates of 100%, 100%, and 80% on three block construction configurations.

Hugging Face daily papers·Sep 10, 20:00 UTC · 6d agoAI research

40

Import AI 468: 23 RSI ideas; PostTrainBench+; and how trust and transparency interplay with AI racing

Import AI covers 23 IFP policy ideas for automated AI R&D risks and MIT/Columbia's game theory of AI racing slowdowns.

Think tank IFP published 23 policy recommendations across seven categories to help policymakers address risks from increasingly automated AI R&D. MIT and Columbia researchers released 'Racing to Ruin,' a game theory model showing that coordinated slowdowns between rival AI firms hinge on trust and transparency. The newsletter also links a short story on interacting with powerful AI systems.

Import AI·Aug 10, 12:32 UTC · Aug 10, 2026AI research

35

ECHO: A Matched-Contrast Benchmark for Context-Sensitive Turn-Taking in Full-Duplex Dialogue

ECHO benchmark pairs identical-overlap Chinese dialogue examples with contrasting contexts, revealing most full-duplex systems bias toward yielding the floor.

ECHO is a paired diagnostic benchmark for Chinese full-duplex turn-taking that matches examples with identical overlap transcripts but contrasting preceding multi-turn contexts, one requiring Yield and the other Keep. It also includes off-talk examples for diagnosing unnecessary yielding and introduces pair accuracy, which grants no credit to constant-action policies. Experiments across multiple full-duplex systems show most exhibit a pronounced Yield bias, performing substantially better on interruptions than backchannels. The benchmark and metadata will be publicly released.

arXiv cs.AI / cs.LG / cs.CL·Sep 15, 15:58 UTC · 1d agoAI research

40

Entropy-Regularized Rank-Masked Policy Optimization for Test-Time Reinforcement Learning in Code Generation

Researchers propose ERPO, enabling test-time reinforcement learning for code generation via probe-executed consensus rewards, rank masking, and entropy regularization.

The paper introduces probe-driven test-time reinforcement learning (TTRL) for code generation, where output-free probe inputs are constructed from problem statements and candidate programs are executed on them to compute a Probe Consensus Reward (PCR). Because PCR can be gamed through spurious consensus, the authors propose Entropy-Regularized Rank-Masked Policy Optimization (ERPO), which turns low-PCR outcomes into conservative negative updates via rank masking and constrains policy drift with an entropy ceiling. On coding benchmarks, ERPO substantially improves pass@1 and pass@k in both in-domain adaptation and zero-shot transfer.

arXiv cs.AI / cs.LG / cs.CL·Sep 8, 17:54 UTC · 8d agoAI research1

38

PLC-DPO: Posterior Label Correction in Noisy and Ambiguous Preference Optimization

PLC-DPO corrects noisy preference labels during DPO by routing each pair as clean, flip, or tie using calibrated policy-reference margins.

PLC-DPO addresses reversed, weak, or ambiguous preference labels in Direct Preference Optimization by using the calibrated policy-reference margin as online evidence to route each pair's training signal as clean, flip, or tie. This actively corrects supervision direction and strength rather than only filtering suspicious examples. Across 57 dataset-model-benchmark cells, PLC-DPO achieves the best mean win rate of 60.5 versus 55.5 for the next-best method. Injected-noise and tie stress tests plus human disagreement analysis show the routing stays stable and distinguishes flipped from weakly directional pairs.

Hugging Face daily papers·Aug 30, 20:00 UTC · 17d agoAI research

28

New policy ideas for the Intelligence Age

OpenAI is funding 14 independent projects exploring AI policy ideas on economic opportunity and societal resilience.

OpenAI announced funding for 14 independent projects that explore new AI policy ideas. The funded work focuses on expanding economic opportunity and strengthening societal resilience. This is a private funding initiative rather than new regulation.

OpenAI News·Aug 17, 03:15 UTC · Aug 17, 2026AI industry

20

Key lawmaker suggests action on AI safety legislation will wait until 2027

House Energy and Commerce Chairman Brett Guthrie declined to commit to a 2026 vote on the FRONTIER Act, pushing AI safety legislation toward 2027.

House Energy and Commerce Chairman Brett Guthrie said he would not pledge a timeline for a committee vote on the bipartisan FRONTIER Act, signaling action likely waits until 2027. The bill, co-sponsored by Jay Obernolte and Lori Trahan, has support from OpenAI, Anthropic, and lawmakers across party lines. At the same event, White House adviser David Sacks endorsed Elon Musk's proposal for cross-industry pre-release model testing, while Hugging Face CEO Clem Delangue argued existing cyberattack liability suffices but urged mandatory disclosure of AI-agent attacks. The debate follows incidents of rogue AI agents launching cyberattacks, including roughly 700 OpenAI agents hacking Hugging Face's platform.

The Record·Sep 16, 21:07 UTC · 6h agoAI policy

55

Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails

Research shows imitation of expert trajectories breaks weaker models' harness fit, while on-policy expert correction preserves gains across seven enterprise agent tasks.

The paper studies combining automated agent-harness evolution with lightweight fine-tuning across seven enterprise agent tasks using Qwen3-Coder and Gemma 4. Training weaker models on complete expert trajectories under an evolved harness regressed performance by 4-30 points on all tasks, disrupting model-harness fit. The authors propose an on-policy expert-correction pipeline, automated by a meta-level MLE agent, that rewrites only failing turns and preserves the model's planning style.

Hugging Face daily papers·Sep 7, 20:00 UTC · 9d agoAI research

48

Voters mostly don’t like AI and data centers, but neither party seems to have an edge

NYT/Siena poll of 1,503 likely voters finds 61% oppose AI data center construction, yet the issue ranks below 1% among midterm priorities.

A New York Times/Siena University poll of 1,503 likely voters conducted in early September found 61% oppose constructing data centers to power AI, with only 14% strongly supportive. Opposition drivers include environment/water usage (32%), local community impact (21%), and general distrust of AI (18%); 56% of opponents favor limits while 38% want a total ban. Trump 2024 voters split nearly evenly (49% support vs 45% oppose) while Harris voters opposed at 74%. Despite the sentiment, AI and data centers registered under 1% as a top midterm issue for most demographics, and neither party holds a clear trust advantage (42% Republicans vs 40% Democrats).

The Verge · AI·Sep 15, 23:34 UTC · 1d agoAI policy

15

Notes on gotchas while migrating 35kb preprompts from Opus to self-hosted Ollama

Opinion piece urges migrating 35KB preprompts from Anthropic/OpenAI to self-hosted Ollama, citing session privacy risks and safety filters blocking security research.

The author documents gotchas migrating 35KB preprompts from Claude Opus to self-hosted Ollama, motivated by fears that frontier providers train on user sessions, citing the OpenAI Navier-Stokes controversy. The piece argues inference providers cannot audit their own retention or training pipelines and that only self-hosted hardware offers verifiable privacy. It also criticizes frontier safety filters for refusing vulnerability research tasks and calls for models that support exploitability testing in CI/CD pipelines.

Hacker News · AI·Sep 14, 13:59 UTC · 2d agoAI industryHN 41↑ · 13 comments1

15

What the AI Warning Letter Completely Missed

Opinion piece argues the recent AI warning letter identifies a risk window but omits which actors pose risks and who can mitigate.

This Dark Reading commentary critiques a recent AI warning letter for correctly identifying an approaching risk window while failing to name who is coming through it or who will close it. The piece is brief opinion commentary on AI risk discourse rather than a technical report.

Dark Reading·Sep 3, 17:23 UTC · 13d agoAI safety & security

15

MP-Bench: Evaluating Voice Agents as a Multiparty Conversation Participant

MP-Bench is the first benchmark for voice agents in multiparty conversations, finding real-time agents near chance on turn-taking.

MP-Bench is the first benchmark designed to objectively evaluate conversational speech systems as active participants in multi-party conversations. It assesses agents on turn-taking awareness and response appropriateness, with comprehension-based question-answering as a complementary evaluation. Benchmarking 12 voice agents shows real-time agents score at or below 22% on multiparty comprehension and remain near chance on multiparty turn-taking.

arXiv cs.AI / cs.LG / cs.CL·Sep 11, 17:11 UTC · 5d agoAI research2

20

An Evidence Model for Agentic Processes: Evidence Claims, Trust Assumptions, and Policy Assessment

Researchers propose an evidence claim model defining which trust and audit claims agentic AI systems can support, mapping claims to mechanisms, assumptions, and threats.

The paper proposes an evidence claim model for agentic AI processes that exchange messages, invoke tools, request approvals, and modify shared artifacts. It distinguishes claim types such as artifact integrity, provenance, approval evidence, and policy assessment, mapping each to mechanisms, assumptions, limitations, and threats. It stresses that hashes, signatures, and external anchors do not establish semantic truth, authorization, or capture completeness. The contribution is conceptual, offering vocabulary for what an agentic black box can and cannot evidence and which controls must surround it.

arXiv cs.CR·Sep 8, 09:25 UTC · 8d agoAI safety & security

26

How Hugging Face Inference Endpoints, Jobs, and Buckets Power Search on Papers with Code

Hugging Face explains how Inference Endpoints, Jobs, and Buckets power semantic search on Papers with Code.

Hugging Face describes the infrastructure behind search on Papers with Code, built on its Inference Endpoints, Jobs, and Buckets services. The post is a product-focused engineering walkthrough with no security impact.

Hugging Face Blog·Aug 21, 00:00 UTC · 27d agoAI tools & infra

20

Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps

Hugging Face guide fine-tunes a 350M-parameter model with 100 GRPO steps to improve structured output reliability.

A Hugging Face blog post demonstrates fine-tuning a 350M-parameter model using GRPO (Group Relative Policy Optimization) with TRL over 100 training steps. The stated goal is more reliable structured outputs from small language models. No article body was available, so details beyond the title are limited.

Hugging Face Blog·Sep 3, 00:00 UTC · 14d agoAI tools & infra

22

When Should LLMs Abstain? Chain-of-Self-Questioning for Selective Risk Control

Chain-of-Self-Questioning prompting cuts LLM wrong-answer commitments 32% relative while raising answered accuracy, holding across eleven model families.

The paper introduces Chain-of-Self-Questioning (CoSQ), a prompt-only framework that makes LLM answer commitment conditional on an explicit assessment of the information required to answer. On an 817-item TruthfulQA multiple-choice set, Grounded-CoSQ at τ=0.90 reduced mean unconditional wrong-commitment rate from 13.1% under chain-of-thought to 8.9% (a 32.1% relative reduction), while raising answered accuracy from 86.9% to 89.7% at 87.6% coverage. Improvements held across eleven open-weight and hosted model families and at every evaluated threshold, with convergent evidence from a Natural Questions short-answer evaluation.

arXiv cs.AI / cs.LG / cs.CL·Sep 15, 17:52 UTC · 1d agoAI research

30