The Groundy archive · Page 27 of 34
Browse Groundy's complete archive of 797 articles on AI, developer tools and infrastructure. Page 27 of 34.
625–648 of 797 articles · Newest first
OpenAI's GPT-5 API features map directly to Cursor's agent loop, revealing a co-design relationship. Coding teams must now evaluate editor-model pairs, not editors alone.
DuckDB queries Parquet files on Hugging Face Hub over HTTPS without downloading them first, turning dataset triage from a multi-gigabyte commitment into a LIMIT 100 query.
A 2026 arXiv preprint argues RLHF's single reward signal destroys the reasoning behind human disagreement, proposing machine theory-of-mind as an alignment foundation.
A June 2026 arXiv analysis traces an AI program through the DoD Software Acquisition Pathway, finding no milestones for model re-validation or data provenance.
The TRAP benchmark finds 13 to 43 percent of web agent tasks can be redirected by persuasive page content, exposing a blind spot in current instruction-hierarchy defenses.
OpenAI's URL provenance filter concedes content inspection is intractable. Agents that mix sensitive data with web access face a structural exfiltration risk.
A June 2026 paper frames the AI agent benchmark gap as a sim-to-real problem, giving eval teams a four-part MDP checklist to challenge vendor claims before live deployment.
Vercel's shift from frontend hosting to agent infrastructure is backed by real products and a $9.3B valuation. Whether per-token billing beats per-seat SaaS remains unproven.
A new framework decomposes patent value into per-feature Shapley credits, but courts have not ruled on whether model output replaces expert testimony in damages and M&A.
A June 2026 paper proposes Data Flow Control, moving agent data safety from prompt-level guardrails to deterministic, auditable SQL query policies enforced outside the model.
LLM agents with formal verification repair 12% more network misconfigurations than base models and are 17% safer, but regress on large topologies, limiting production use.
Self-evolving AI agents drift without checkpoints: 94% of reviewers miss agent sabotage, safety hardening does not transfer across domains, and stale memory degrades tasks.
A 2026 analysis of the bot comment archive from a halted Reddit experiment catalogs fabricated identities and bias triggers, but early shutdown leaves harm unmeasurable.
kapa.ai's data shows indexing image captions at ingestion adds 1-6% query overhead versus 27-51% for raw query-time vision, shifting recall risk to caption fidelity.
A facility-level audit of 403 US hyperscale centers finds 545 gCO2/kWh, 48% above the grid average. Siting in fossil-heavy regions, not PPAs, determines actual emissions.
LLM decode is memory-bandwidth-bound, not capacity-bound. A 70B model on the DGX Spark's 273 GB/s hits roughly 2.7 tok/s. Count GB/s, not GB, when sizing inference hardware.
A single-query attack turns safety-trained LLMs' own refusal reasoning against them. Across 30 models, better safety judgment correlated with higher exploit rates, not lower.
XML Signature Wrapping attacks on SAML keep recurring because the gap between validation and processing is structural. Edge WAF rules are a delaying tactic, not a fix.
MIDI tests idiom comprehension across 18 languages and finds LLMs rely on memorization over reasoning, with the sharpest failures falling on low-resource communities.
A new study claims LLMs write 'appropriate' research titles, but the evidence rests on similarity metrics that measure pattern matching, not whether titles actually serve.
dstack-capsule binds pod identity into Intel TDX hardware quotes, enabling multi-pod confidential VMs without the per-VM density tax of Confidential Containers.
Standard concept bottleneck model benchmarks confound genuine concept learning with dataset shortcuts. Synthetic benchmarks from Skirzynski et al. expose the gap.
The CHARM paper shows per-step grounding checks in multi-hop RAG miss over 80% of cascaded errors, where one fabricated retrieval compounds across reasoning hops.
CVE-2025-46332 exposed flag names, rollout conditions, and security kill switches via Vercel's discovery endpoint, making operational metadata into reconnaissance material.