The AI-news segment of Hard Fork connects lab security, obligations left behind by talent deals, and the coordination work household assistants promise to handle.

Yuning Bian and colleagues compare document splitting methods across 120 adjacent Wikipedia revision pairs. A small edit can shift boundaries and force a retrieval system to embed unchanged text again. On the 116 pairs all methods process …

Abhishek Barali and Aayush Paudel test what happens when an extraction schema requires more answers than a document contains. Models often fill the gap with genuine text of the wrong kind. Across eight downloadable models and 558 documents, …

A review of 1,368 reported AI incidents motivates tests of permissions, human review and monitoring alongside model accuracy. The findings are retrospective, not proof of prevention.

Air Canada’s Nathanael Weill starts with known answers, generates test inputs around them and checks results in code. Synthetic tests speed iteration but do not authorize release.

Tone-Bench tests whether models can write in one person’s voice. Louis-François Bouchard explains its rubrics, judge bias and the cost of changing a test.

An audit finds hidden interface requirements in SWE-Milestone. They distort some results, while removing affected tests largely preserves the overall ranking.

Cole Medin argues that better agent workflows should buy longer stretches of autonomy and fewer interruptions, rather than another session to supervise.

A six-stage creator workflow built around reference libraries, reusable skills, editing tools and human approval—not a promise of automatic audience growth.

Jens Lehmann and Sahar Vahdati let models build and twice revise their own harnesses: the software that runs their tools, files and commands. Improvements on familiar development games often overstate gains elsewhere. The study follows 27 …

MemPABench asks whether an assistant remembers how a person wants help, including when to confirm, ask questions or initiate contact. It measures the preference selected and the behavior actually delivered. Four simulated users have …

Text-only editing damaged some correct transcripts. Adding audio reduced that damage for Gemini, but also reduced repairs to its own drafts.

Scheduled check-ins and useful work artifacts stand out; browser handoff and access to local skills remain obstacles in Yang's early-access test.

Match engineering rigor to failure consequences, dependencies and replacement cost—and make the rigorous path cheaper for humans and coding agents.

Directable voices, persistent AI worlds and 3D reconstruction, with the hardware limits and demo-only claims kept in view.

Hirundo reports that naming a different customer changes generated-code security. The evidence is narrower than the backdoor rhetoric.

Yikai Zhao and colleagues test whether document conversion preserves the requirements an author marked for attention. Words can survive extraction while color and highlighting disappear. Their task asks models to recover specially marked …

Maddy Zhang's five-part review framework covers secrets, data flow, dependencies, agent permissions and human accountability. Passing tests covers only part of the risk.

Zedong Peng compares extracted memories with coding agents that search raw conversation history. The study includes store-construction costs, which can outweigh cheaper answers later. It tests seven methods with three models on one LoCoMo …

Peter Yang's video-short skill combines an edit plan, generated visuals and frame-specific feedback. His demo still needs several rounds of human review.

Google opens its PyTorch software for its own chips Google releases TorchTPU for running PyTorch on its chips — The release makes Google’s rented accelerators easier to test with the model libraries and serving software you already use. The …

IBM's Security Intelligence panel connects AI-assisted attacks and vulnerability discovery to continuous patching, agent identity controls and human remediation capacity.

GEICO's Ford Prior brings decision makers and agent fleets into one workflow, pairing legacy-code research with human approvals, reusable diagrams and explicit cost controls.

Vanguard's cash-ledger migration pairs AI-queryable legacy code with daily data checks, migration documentation and visibility into producers and consumers.

MiniMax's Morgan Suo connects faster model internals to the cost, accuracy and delivery time of a research report an analyst can actually use.

Thomas Hikaru Clark, Matias B. Martin and Kyle Mahowald examine why AI prose can feel difficult despite being grammatically valid. Their study finds features associated with reading difficulty across model-generated conversations, coding …

Oliver Kennon explains how serving software, reliability and security change the cost of useful AI output, even on the same GPU.

A poisoned model kept passing clean tool-use tests while a trigger phrase made a coding agent send dummy credentials to an outside server.

Monzo's Agent Chip grew from a Slack reaction into a reusable internal service. Its speakers explain the platform, permissions and sandboxes behind that growth.

A study of ChatGPT, Claude and Gemini shows how saved malicious instructions can affect later chats. Memory injection means storing an instruction that returns as context in future sessions. The authors tested ChatGPT Instant 5.3, Claude …

Invented statistics acquired real citations. A source link can preserve a falsehood as easily as it can verify a fact.

Cole Medin demonstrates two feedback loops for production agents: learn from human corrections, then propose trace-backed changes to instructions and skills.

Jeff Crume uses Jev to explain why automated decisions need independent evidence, audit trails and checks for subtle changes in behavior.

Single-block profiling finds that attention software leaves newer workstation GPUs underused, while proposed fixes remain a systems research agenda.

Redact found instructions that withheld public escalation routes from chatbot answers. Whether any bot followed them remains untested.

DeepSeek opens its model-serving software DeepSeek publishes the software that runs its model — Developers can inspect and adapt DeepInfer to run DeepSeek’s Flash model on supported server hardware. The Apache-2.0 code supports …

Al Chen of Cisco explains how exploit tests, verified restores and privacy checks help banks move AI into production. The bottleneck also includes image approvals, GPU procurement and internal ownership.

Moshi founder Anekha Sokhal argues that AI should gather and check market information while investors keep the decision. She describes a six-agent system, traceable claims and the hidden cost of losing apprenticeship.

Anti-Persistence saves execution records and rebuilds advice for each task, reducing one route by which an agent can keep repeating a bad lesson.

Momento's Khawaja Shams explains realistic coding tasks, leaked answers, repeated trials, and the traces that make agent scores worth trusting.

AI can turn shared knowledge into private wealth while overwhelming the reviewers and maintainers who keep it alive.

A small learned controller compares proposed actions with user authority and an alternative, with important differences between classification and reversible execution tests.

VerifierBench tests whether agents can write executable task checkers. An agent is a model running a multi-step task with tools. A checker must recognize different valid solutions, including ones its author never chose. The authors report a …

Benj Edwards makes songs and games for an audience of one. Their private value can be real while the consequences for creative workers remain unresolved.

Request-level logs make coding agents easier to debug. The HN discussion asks whether that requires putting another service in every request's path.

Brooke Lacey shows Claire Vo how she uses Claude Code, FFmpeg, and Whisper to turn phone footage into short-form videos, learn her delivery, and mine audience comments.

Agent frameworks & tooling Munin lets research groups question papers on their own servers paper notes → — Its released software includes tests you can rerun after changing the model or paper collection. An agent is a model running a …

Martin Keen explains the goals, triggers, context, tools and human checks around an agent's act–observe–adjust loop.

Store the order and prerequisites of successful work, then retrieve compact instructions instead of loading every skill.

A two-model GEPA and SkillOpt study separates answer correctness, formatting, server failures, waiting time and token use.