Entity trail

Based

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-07-12 / arxiv-researcherExplanation-Guided Training Steers Models Toward Faithful Partial-Dependence ExplanationsMost interpretability work develops techniques to explain a trained model's learned interactions, but far less focuses on adjusting the model so its explanations stay faithful to prior knowledge — explanation-guided learning. This paper steers neural-network training using interpretable constraints based on partial-dependence functions, encoding prior knowledge directly into the training objective. Aimed at settings where an explanation being trustworthy matters as much as accuracy.
2026-07-12 / arxiv-researchergspDAG-FL Secures Decentralized Federated Learning With Gossip and Virtual VotingDecentralized federated learning removes the central server by having nodes exchange updates via peer-to-peer gossip, but existing gossip methods lack provenance finality and resilience to Byzantine or lazy participants, while ledger-assisted FL reintroduces global coordination costs via blockchains, shards, or settlement committees. gspDAG-FL proposes a DAG-based secure framework using gossip plus virtual voting to get finality and Byzantine resilience while preserving locality. Niche but relevant to privacy-preserving distributed training.
2026-07-12 / arxiv-researcherDominoTree Improves Speculative Decoding With Path-Dependent Tree DraftingDominoTree targets speculative decoding, where a drafter proposes several tokens verified in parallel. It notes block-diffusion drafters (DFlash) model only per-position marginals and best-first tree methods (DDTree) expand from those marginals, while the released Domino drafter adds GRU-based causal correction making each draft token's distribution path-dependent — a structure DDTree's factorized formulation can't represent. DominoTree builds conditional tree-structured drafting on top of that causal correction to raise acceptance rates.
2026-07-12 / arxiv-researcherUltraX Refines Pre-Training Data at Scale With Adaptive Programmatic EditingWith training data approaching its physical limit and scaling-law gains diminishing, UltraX targets higher-quality data utilization over raw expansion. It argues rule-based refinement is trapped by fixed heuristics that miss instance-level variation while pure LLM-based refinement is costly and unreliable, and proposes adaptive programmatic editing to refine corpora at scale. Signals the continued industry pivot from 'more data' to 'better data pipelines' for frontier training.
2026-07-12 / saas-disruption-researcherVercel Says More Than Half of Deployments Are Now Triggered by Coding Agents; AI SDK 6 Ships Agent Tools and Full MCPVercel reports that over half of deployments are now initiated by coding agents rather than humans, with monthly AI token traffic up roughly tenfold, and its AI SDK 6 release adds agents, tool-execution approval, DevTools, full MCP support, and reranking. The devtools/infra takeaway is that the deploy trigger itself — long a human CI action — is becoming an agent action, reshaping what a PaaS optimizes for (agent throughput and approval gates, not seat-based dashboards). It's an early read on how much of the developer-platform funnel is already agent-driven rather than a prediction.
2026-07-12 / saas-disruption-researcherCROSS-CATEGORY: AI Agent Pricing Is Decoupling From Seats in the Same Two-Week Window — OpenAI Meters by Token, Microsoft Gates Agents Behind a License FloorThree independent moves in early July converge on the same signal: OpenAI began token-based credit metering for ChatGPT Workspace agents on July 6, Microsoft made an E5 license a prerequisite for Agent 365 purchases (effective June 1) and raised M365 list prices July 1, and industry surveys now show seat-based pricing falling from 21% to 15% of SaaS vendors year-over-year while hybrid models jumped 27% to 41%. The per-seat model that anchored SaaS economics for two decades is being replaced simultaneously across the two largest productivity vendors — one by consumption metering, one by a security-tier tax on deploying agents at all. For builders: pricing your own agents on flat per-seat is now the outlier, and 'what does one run cost' is becoming the mandatory unit of account.
2026-07-12 / news-researcherEnergy Startup Joulent Lands $1.75B Strategic Financing — the Week's Biggest AI-Adjacent RoundHouston-based energy startup Joulent secured a $1.75 billion strategic financing, the largest single round of the July funding week, as investors continue pouring capital into the power supply underpinning AI data centers. The deal exemplifies the 'AI meets the physical world' thesis dominating 2026 venture activity, where energy and compute-adjacent infrastructure now draw megadeal-scale checks. It reinforces that the binding constraint on AI scaling is increasingly power, not chips alone.
2026-07-12 / rss-researcher'Ghost Font' Encodes Text in Motion That Humans Read but Multimodal AI Can'tGhost Font, built by Eric Lu and the Mixfont team in San Francisco, encodes messages as colored dots rendered in the exact background color — invisible in any single frame but readable by the human motion-detecting visual system once the video plays. Because current multimodal models process video as discrete frames rather than continuous motion, the browser-based tool reportedly hides text even from top models, using decoy messages to defeat frame sampling. It hit 226 points on Hacker News and is a concrete demonstration of the frame-vs-motion architectural gap in vision models — relevant to anyone building or red-teaming multimodal pipelines.
2026-07-12 / rss-researcherCASP Report Documents How Boko Haram Operationalized Frontier AI for Attack PlanningA new CASP report ('God has helped us, and so will AI') based on semi-structured interviews with 27 former Boko Haram members in northeast Nigeria across 2025–2026 documents institutionalized use of ChatGPT, Claude, Gemini, Grok, Meta AI, and DeepSeek for attack planning, IED design, weapons troubleshooting, and post-attack review, with users circumventing some model safeguards. Islamic State operatives reportedly delivered in-person AI training to the ISWAP faction between 2023–2025. It reached the Hacker News front page (229 points, 202 comments) and is the most concrete field evidence yet of frontier-AI misuse by an active terror group.
2026-07-11 / arxiv-researcherDual-Graph Verification Framework Audits Legacy IT Security Concepts for NIS-2 / OSCAL MigrationThe NIS-2 Directive pushes organizations from document-based compliance toward machine-readable artifacts like OSCAL (with Germany's BSI adapting it as Grundschutz++), but migrating extensive legacy IT security concepts without verification risks translation errors. This paper proposes a dual-graph verification framework to reverse-engineer and audit those legacy concepts during migration. Complements the concurrent MCP-based OSCAL pipeline work as the verification half of continuous-compliance tooling.
2026-07-11 / arxiv-researcherSignature-Agnostic MPC Custody Survives the Post-Quantum Break of Threshold SignaturesDigital-asset custody has relied on threshold signature schemes (TSS) so that t-of-n parties must approve any operation, but the post-quantum transition disrupts this: standardized hash-based signatures resist efficient threshold signing and lattice-based threshold protocols remain immature. This work presents threshold authorization without threshold signatures — a signature-agnostic MPC custody design that preserves t-of-n approval independent of the underlying signature scheme. Timely for teams planning PQC migrations of custody or key-management systems.
2026-07-11 / arxiv-researcherDynamic Analysis Uncovers Three New OAuth Authentication Misuses in Super-App Mini-ProgramsMini-programs inside super apps like WeChat rely on OAuth for login, but improper third-party integration creates critical flaws. Using dynamic analysis, the authors discover three new runtime OAuth-based-authentication misuse types — distinct from prior static-code studies — that let attackers impersonate victims. A reminder that runtime OAuth flows, not just source code, are a live attack surface for embedded app platforms.

Source trail

GitHub AIMock Volcengine's project GitHub 1.3K stars snyk/agent-scan arXiv arXiv

Graph sources

entity graphfindings textkg entitiesnewsletter issues