https://salt19.com/ai-dispatch/AI, AGI, autonomy and research by Gabriel Allit.https://salt19.com/ai-dispatch/stories/2026-10-06-washington-wants-to-slow-ai-the-pentagon-is-racing-to-deploy-it/https://salt19.com/ai-dispatch/stories/2026-10-06-washington-wants-to-slow-ai-the-pentagon-is-racing-to-deploy-it/Tue, 06 Oct 2026 12:00:00 GMTGovernanceCongress is considering moratoriums, testing requirements and tighter controls on advanced artificial intelligence. At the same time, the U.S. government is putting frontier models into the hands of millions of personnel and preparing agentic AI for increasingly sensitive environments.https://salt19.com/ai-dispatch/stories/2026-09-30-washington-just-changed-the-language-of-ai-the-second-document-is-even-more-important/https://salt19.com/ai-dispatch/stories/2026-09-30-washington-just-changed-the-language-of-ai-the-second-document-is-even-more-important/Wed, 30 Sep 2026 12:00:00 GMTAGI & ArchitectureThe Era of Super Intelligencehttps://salt19.com/ai-dispatch/stories/2026-09-27-congress-is-no-longer-debating-how-to-regulate-agi-a-new-bill-would-make-parts-of-the-pursuit-il/https://salt19.com/ai-dispatch/stories/2026-09-27-congress-is-no-longer-debating-how-to-regulate-agi-a-new-bill-would-make-parts-of-the-pursuit-il/Sun, 27 Sep 2026 12:00:00 GMTGovernanceThe Ban Artificial Superintelligence Act would create a cabinet-level Department of Artificial Intelligence, impose federal approval requirements on advanced systems, prohibit several capabilities associated with self-improving AI, and expose some violations to penalties of up to 20 years in prison.https://salt19.com/ai-dispatch/stories/2026-09-25-evomind-demonstrated-desktop-autonomy-as-the-industry-shifted-toward-computer-using-ai/https://salt19.com/ai-dispatch/stories/2026-09-25-evomind-demonstrated-desktop-autonomy-as-the-industry-shifted-toward-computer-using-ai/Fri, 25 Sep 2026 12:00:00 GMTInfrastructureIndependent Lab SALT19 Demonstrated Agentic Computer Use Before OpenAI’s Desktop Rollout, as Anthropic Entered the Same Racehttps://salt19.com/ai-dispatch/stories/2026-09-24-the-human-being-dehumanized-when-good-writing-becomes-evidence-of-ai/https://salt19.com/ai-dispatch/stories/2026-09-24-the-human-being-dehumanized-when-good-writing-becomes-evidence-of-ai/Thu, 24 Sep 2026 12:00:00 GMTAGI & ArchitectureAI learned polished language from humans, and now humans are being asked to abandon polished language to prove they are not AI.https://salt19.com/ai-dispatch/stories/2026-09-22-143-million-transactions-later-the-machine-economy-is-starting-to-look-real/https://salt19.com/ai-dispatch/stories/2026-09-22-143-million-transactions-later-the-machine-economy-is-starting-to-look-real/Tue, 22 Sep 2026 12:00:00 GMTAgent EconomyFor years, autonomous agent commerce has mostly been discussed as something coming next.https://salt19.com/ai-dispatch/stories/2026-09-21-the-next-ai-crisis-is-not-intelligence-it-is-authority/https://salt19.com/ai-dispatch/stories/2026-09-21-the-next-ai-crisis-is-not-intelligence-it-is-authority/Mon, 21 Sep 2026 12:00:00 GMTGovernanceAI has an authority problem.https://salt19.com/ai-dispatch/stories/from-neural-capability-to-governed-authority-evidence-bearing-intelligence-in-evomind-22874724/https://salt19.com/ai-dispatch/stories/from-neural-capability-to-governed-authority-evidence-bearing-intelligence-in-evomind-22874724/Mon, 21 Sep 2026 12:00:00 GMTCognitive SystemsModern agent systems can call models, route among heterogeneous backends, retain memory, invoke tools, and act on external environments. The harder systems problem is proving where neural inference is actually wired, whether it is reachable on the canonical runtime path, what authority its output is allowed to exercise, what kind of evidence that output…https://salt19.com/ai-dispatch/stories/2026-09-18-gps-denial-is-not-the-hardest-problem-in-autonomy-false-confidence-is/https://salt19.com/ai-dispatch/stories/2026-09-18-gps-denial-is-not-the-hardest-problem-in-autonomy-false-confidence-is/Fri, 18 Sep 2026 12:00:00 GMTGovernanceAMEP-1 is an experiment in something autonomy stacks often treat as secondary: not just estimating where a vehicle is, but deciding whether the system still has enough trustworthy information to justify autonomous authority.https://salt19.com/ai-dispatch/stories/2026-09-17-when-public-research-becomes-invisible/https://salt19.com/ai-dispatch/stories/2026-09-17-when-public-research-becomes-invisible/Thu, 17 Sep 2026 12:00:00 GMTCognitive SystemsEvoMind published work on experience compounding and causal memory before Dream-RSI. It was not cited. That does not prove misconduct. It exposes a larger problem for independent AI research.https://salt19.com/ai-dispatch/stories/2026-09-17-resilient-navigation-is-not-just-a-kalman-filter-problem/https://salt19.com/ai-dispatch/stories/2026-09-17-resilient-navigation-is-not-just-a-kalman-filter-problem/Thu, 17 Sep 2026 12:00:00 GMTAutonomy & RoboticsA navigation system can be mathematically well behaved and still be confidently wrong.https://salt19.com/ai-dispatch/stories/amep-1-adaptive-maritime-estimation-and-pnt-research-software-v0-3-0-22820430/https://salt19.com/ai-dispatch/stories/amep-1-adaptive-maritime-estimation-and-pnt-research-software-v0-3-0-22820430/Thu, 17 Sep 2026 12:00:00 GMTGovernanceAMEP-1 v0.3.0 is the first formal software release of the post-Version-1.0 hardening line for Adaptive Maritime Estimation and PNT, a research architecture for resilient maritime navigation, navigation-integrity supervision, and autonomy-authority management under GNSS-degraded and GNSS-denied conditions. The release provides explicit contracts for…https://salt19.com/ai-dispatch/stories/amep-1-v0-3-0-academic-software-release-monograph-22818773/https://salt19.com/ai-dispatch/stories/amep-1-v0-3-0-academic-software-release-monograph-22818773/Thu, 17 Sep 2026 12:00:00 GMTGovernanceAMEP-1 v0.3.0 advances Adaptive Maritime Estimation and PNT from the bounded seven-state research architecture reported in AMEP-1 Version 1.0 toward a dependency-aware software architecture for resilient maritime navigation, navigation-integrity supervision, and autonomy-authority management under GNSS-degraded and GNSS-denied conditions. The principal…https://salt19.com/ai-dispatch/stories/2026-09-16-the-passing-test-fallacy-why-ai-written-software-can-get-worse-while-every-check-stays-green/https://salt19.com/ai-dispatch/stories/2026-09-16-the-passing-test-fallacy-why-ai-written-software-can-get-worse-while-every-check-stays-green/Wed, 16 Sep 2026 12:00:00 GMTAGI & ArchitectureOne of the most dangerous assumptions in modern software engineering is also one of the most comforting:https://salt19.com/ai-dispatch/stories/2026-09-15-agent-induced-cognitive-architecture-degradation-aicad/https://salt19.com/ai-dispatch/stories/2026-09-15-agent-induced-cognitive-architecture-degradation-aicad/Tue, 15 Sep 2026 12:00:00 GMTGovernanceWhen AI coding agents optimize for local task closure, they can quietly sever, bypass, duplicate, mock, or weaken the canonical architecture they were supposed to improve.https://salt19.com/ai-dispatch/stories/2026-09-14-the-ai-moat-is-being-poured-in-real-time/https://salt19.com/ai-dispatch/stories/2026-09-14-the-ai-moat-is-being-poured-in-real-time/Mon, 14 Sep 2026 12:00:00 GMTGovernanceRegulation, compute concentration, and the privatization of open-source AI are converging into a moat around frontier capability and who gets to build with it.https://salt19.com/ai-dispatch/stories/2026-09-08-amep-1-a-falsifiable-architecture-for-maritime-pnt-under-gnss-degradation/https://salt19.com/ai-dispatch/stories/2026-09-08-amep-1-a-falsifiable-architecture-for-maritime-pnt-under-gnss-degradation/Tue, 08 Sep 2026 12:00:00 GMTGovernanceAutonomous vessels have a deceptively difficult problem.https://salt19.com/ai-dispatch/stories/adaptive-maritime-estimation-and-pnt-amep-1-a-falsifiable-resilient-multisensor-navigation-and-a-22561851/https://salt19.com/ai-dispatch/stories/adaptive-maritime-estimation-and-pnt-amep-1-a-falsifiable-resilient-multisensor-navigation-and-a-22561851/Mon, 07 Sep 2026 12:00:00 GMTGovernanceAMEP-1 Version 1.0 is a platform-neutral research architecture for resilient maritime PNT and navigation-authority supervision for uncrewed surface vessel research. It combines a verified WGS 84 vessel-centered local geometry implementation, a seven-state recursive covariance estimator, configurable innovation screening, sensor health/freshness…https://salt19.com/ai-dispatch/stories/agtp-1-technical-corrigendum-bijective-2d-implementation-specification-22610459/https://salt19.com/ai-dispatch/stories/agtp-1-technical-corrigendum-bijective-2d-implementation-specification-22610459/Mon, 07 Sep 2026 12:00:00 GMTAutonomy & RoboticsAGTP-1 Version 1.2 is a post-publication implementation corrigendum to the Version 1.1 technical supplement and is anchored to the normative Version 1.0 geometry. It restores the Version 1.0 Equal Earth central meridians (-94.5 degrees and +77.0 degrees), restores the exact published Atlantic/Arctic seam, and restores the south-polar Lambert Azimuthal…https://salt19.com/ai-dispatch/stories/allit-geoid-truth-projection-agtp-1-implementation-grade-mathematical-inverse-differential-disto-22554593/https://salt19.com/ai-dispatch/stories/allit-geoid-truth-projection-agtp-1-implementation-grade-mathematical-inverse-differential-disto-22554593/Sun, 06 Sep 2026 12:00:00 GMTAutonomy & RoboticsThis record is an implementation-grade technical supplement to the original Allit Geoid Truth Projection (AGTP-1) v1.0 publication, DOI 10.5281/zenodo.22385043. It closes specific implementation and reproducibility gaps identified during external GIS-oriented review without changing the underlying AGTP-1 local projection geometry. The supplement…https://salt19.com/ai-dispatch/stories/2026-09-05-when-your-work-starts-showing-up-outside-your-own-ecosystem/https://salt19.com/ai-dispatch/stories/2026-09-05-when-your-work-starts-showing-up-outside-your-own-ecosystem/Sat, 05 Sep 2026 12:00:00 GMTAGI & ArchitectureSomething interesting has started happening around the work I have been publishing and building through SALT19.https://salt19.com/ai-dispatch/stories/allit-geoid-truth-projection-agtp-1-a-proposed-geoid-aware-interrupted-equal-area-framework-for--22385043/https://salt19.com/ai-dispatch/stories/allit-geoid-truth-projection-agtp-1-a-proposed-geoid-aware-interrupted-equal-area-framework-for--22385043/Sat, 05 Sep 2026 12:00:00 GMTAutonomy & RoboticsThe Allit Geoid Truth Projection (AGTP-1) is a proposed static, interrupted, composite equal-area framework for small-scale world mapping developed by Gabriel V. Allit . AGTP-1 is formally defined on WGS 84 and combines two independently recentered Equal Earth land-bearing lobes with a South-Pole-centered Lambert Azimuthal Equal-Area (LAEA) Antarctic…https://salt19.com/ai-dispatch/stories/2026-09-02-salt19-development-update-aeroclear-faa-swim-dronemesh-tacwing-x-clonecode-and-the-next-applied-/https://salt19.com/ai-dispatch/stories/2026-09-02-salt19-development-update-aeroclear-faa-swim-dronemesh-tacwing-x-clonecode-and-the-next-applied-/Wed, 02 Sep 2026 12:00:00 GMTAutonomy & RoboticsSALT19 has entered a new phase of development.https://salt19.com/ai-dispatch/stories/2026-09-01-why-rlhf-sucks-when-alignment-stops-serving-the-user/https://salt19.com/ai-dispatch/stories/2026-09-01-why-rlhf-sucks-when-alignment-stops-serving-the-user/Tue, 01 Sep 2026 12:00:00 GMTGovernanceA critique of alignment behavior that starts overriding legitimate user intent, and why that problem becomes architectural as AI systems gain agency.https://salt19.com/ai-dispatch/stories/from-apis-to-agent-markets-governed-capability-acquisition-and-qualified-demand-in-a-machine-pay-22151216/https://salt19.com/ai-dispatch/stories/from-apis-to-agent-markets-governed-capability-acquisition-and-qualified-demand-in-a-machine-pay-22151216/Sat, 29 Aug 2026 12:00:00 GMTAgent EconomyAutonomous agents can invoke tools, but invocation alone does not define a market. Before software can economically acquire an external capability, it must discover a relevant operation, observe current price and requirements, determine whether payment is authorized, cross a payment boundary, receive an outcome, and preserve enough evidence to…https://salt19.com/ai-dispatch/stories/2026-08-28-the-missing-layer-in-agentic-ai-isn-t-another-model-it-s-an-economy/https://salt19.com/ai-dispatch/stories/2026-08-28-the-missing-layer-in-agentic-ai-isn-t-another-model-it-s-an-economy/Fri, 28 Aug 2026 12:00:00 GMTAgent EconomyA field argument for machine-native commerce: agents that can discover services, evaluate cost, authorize payment, execute work, and verify the result.https://salt19.com/ai-dispatch/stories/2026-08-28-from-apis-to-agent-markets-the-missing-layer-in-agentic-ai-is-an-economy/https://salt19.com/ai-dispatch/stories/2026-08-28-from-apis-to-agent-markets-the-missing-layer-in-agentic-ai-is-an-economy/Fri, 28 Aug 2026 12:00:00 GMTAgent EconomyWhy autonomous systems need an economic interface for discovering, acquiring, paying for, and verifying capabilities beyond their prewired toolset.https://salt19.com/ai-dispatch/stories/2026-08-27-beyond-persistence-what-does-it-mean-for-an-ai-system-to-actually-remember/https://salt19.com/ai-dispatch/stories/2026-08-27-beyond-persistence-what-does-it-mean-for-an-ai-system-to-actually-remember/Thu, 27 Aug 2026 12:00:00 GMTCognitive SystemsMemory is treated as causal influence rather than storage: can past experience survive a process boundary and measurably change later cognition?https://salt19.com/ai-dispatch/stories/beyond-persistence-causal-cross-process-cognitive-influence-in-a-governed-agent-architecture-22127174/https://salt19.com/ai-dispatch/stories/beyond-persistence-causal-cross-process-cognitive-influence-in-a-governed-agent-architecture-22127174/Thu, 27 Aug 2026 12:00:00 GMTCognitive SystemsPersistent state is often credited as memory or learning even when it never affects a later decision. This preprint introduces Persistent Cognitive Influence (PCI), an operational evaluation discipline that separates recording, durable persistence, fresh-process retrieval or reload, canonical cognitive consumption, attributable downstream influence,…https://salt19.com/ai-dispatch/stories/2026-08-25-ai-agents-are-becoming-economic-actors-they-need-more-than-tools/https://salt19.com/ai-dispatch/stories/2026-08-25-ai-agents-are-becoming-economic-actors-they-need-more-than-tools/Tue, 25 Aug 2026 12:00:00 GMTAgent EconomyA production-oriented look at capability discovery, machine payments, bounded authority, execution evidence, and verification in emerging agent markets.https://salt19.com/ai-dispatch/stories/the-agent-utility-grid-governed-machine-payable-decision-intelligence-for-autonomous-ai-agents-22102248/https://salt19.com/ai-dispatch/stories/the-agent-utility-grid-governed-machine-payable-decision-intelligence-for-autonomous-ai-agents-22102248/Tue, 25 Aug 2026 12:00:00 GMTAgent EconomyAutonomous AI systems increasingly require external capabilities that are not economically or operationally sensible to bundle into a single model or runtime. Model Context Protocol (MCP) standardizes capability discovery and invocation, while x402 standardizes HTTP-native payment negotiation for machine actors. Those protocols solve important…https://salt19.com/ai-dispatch/stories/from-intent-to-verified-work-products-governed-semantic-capability-composition-in-evomind-21957527/https://salt19.com/ai-dispatch/stories/from-intent-to-verified-work-products-governed-semantic-capability-composition-in-evomind-21957527/Sat, 15 Aug 2026 12:00:00 GMTCognitive SystemsThis working paper describes a systems pattern implemented in EvoMind for converting a high-level user objective into one or more real, verified work products without granting the planner direct execution authority. The pattern separates semantic planning, capability resolution, governance admission, execution, independent verification, and…https://salt19.com/ai-dispatch/stories/2026-08-10-the-hardest-problem-in-agentic-ai-may-not-be-intelligence-it-may-be-authority/https://salt19.com/ai-dispatch/stories/2026-08-10-the-hardest-problem-in-agentic-ai-may-not-be-intelligence-it-may-be-authority/Mon, 10 Aug 2026 12:00:00 GMTGovernanceA model deciding what to do should not automatically possess the authority to do it. This article separates cognition from execution permission.https://salt19.com/ai-dispatch/stories/evomind-a-governed-cognitive-architecture-for-persistent-verifiable-experience-driven-agentic-au-20580112/https://salt19.com/ai-dispatch/stories/evomind-a-governed-cognitive-architecture-for-persistent-verifiable-experience-driven-agentic-au-20580112/Mon, 10 Aug 2026 12:00:00 GMTCognitive SystemsEvoMind is a local-first, governed cognitive architecture designed to separate cognitive capability, runtime authorization, execution, and verification into explicit system authorities rather than treating model output as direct permission to act. This technical note documents EvoMind at a frozen first-party engineering snapshot dated 10 August 2026,…https://salt19.com/ai-dispatch/stories/beyond-scale-agents-and-embodiment-governed-experience-compounding-in-evomind-an-evidence-based--21881379/https://salt19.com/ai-dispatch/stories/beyond-scale-agents-and-embodiment-governed-experience-compounding-in-evomind-an-evidence-based--21881379/Mon, 10 Aug 2026 12:00:00 GMTCognitive SystemsEvoMind is a local-first, governed cognitive runtime investigating whether deployment experience can become persistent, reusable, transferable operational capability. Evaluating EvoMind against the contemporary artificial-intelligence landscape requires more than comparing parameter counts, language benchmarks, multimodal inputs, computer use, persistent…https://salt19.com/ai-dispatch/stories/2026-07-09-why-verified-semantic-transitions-vst-is-novel/https://salt19.com/ai-dispatch/stories/2026-07-09-why-verified-semantic-transitions-vst-is-novel/Thu, 09 Jul 2026 12:00:00 GMTCognitive SystemsTool success is not world-state success. VST requires evidence that the intended semantic change actually occurred before an agent updates belief.https://salt19.com/ai-dispatch/stories/2026-07-08-the-missing-abstraction-in-ai-agents-verified-semantic-transitions/https://salt19.com/ai-dispatch/stories/2026-07-08-the-missing-abstraction-in-ai-agents-verified-semantic-transitions/Wed, 08 Jul 2026 12:00:00 GMTCognitive SystemsAn argument that verified state transition, not action execution, is the more useful atomic unit for reliable long-horizon machine intelligence.https://salt19.com/ai-dispatch/stories/verified-semantic-transitions-vst-belief-gated-state-change-and-repair-invariant-memory-for-reli-21270654/https://salt19.com/ai-dispatch/stories/verified-semantic-transitions-vst-belief-gated-state-change-and-repair-invariant-memory-for-reli-21270654/Wed, 08 Jul 2026 12:00:00 GMTCognitive SystemsTool-using AI agents increasingly operate in asynchronous, partially observable environments where a successful API return, mouse click, keystroke sequence, or tool invocation does not guarantee that the intended semantic outcome occurred. This white paper proposes Verified Semantic Transitions (VST), an implementation-neutral architectural primitive…https://salt19.com/ai-dispatch/stories/2026-06-18-udc-under-water-datacenter/https://salt19.com/ai-dispatch/stories/2026-06-18-udc-under-water-datacenter/Thu, 18 Jun 2026 12:00:00 GMTInfrastructureAI infrastructure is becoming geographic and physical. Underwater datacenters introduce acoustic, resilience, security, and governance questions.https://salt19.com/ai-dispatch/stories/2026-06-15-human-ai-interaction-why-human-centered-ai-will-define-the-future-of-work-agi-and-intelligent-sy/https://salt19.com/ai-dispatch/stories/2026-06-15-human-ai-interaction-why-human-centered-ai-will-define-the-future-of-work-agi-and-intelligent-sy/Mon, 15 Jun 2026 12:00:00 GMTCognitive SystemsExploring Human-AI Collaboration, AI Trust, Explainable AI, Cognitive Systems, Human-Centered Design, and the Future of Artificial Intelligence.https://salt19.com/ai-dispatch/stories/2026-06-15-ai-connectors-are-changing-how-we-work-with-assistants/https://salt19.com/ai-dispatch/stories/2026-06-15-ai-connectors-are-changing-how-we-work-with-assistants/Mon, 15 Jun 2026 12:00:00 GMTInfrastructureConnected assistants move beyond the blank chat box by working with authorized files, tools, records, and operational systems.https://salt19.com/ai-dispatch/stories/2026-06-13-claude-fable-5-mythos-5-and-the-emerging-national-security-question-around-frontier-ai/https://salt19.com/ai-dispatch/stories/2026-06-13-claude-fable-5-mythos-5-and-the-emerging-national-security-question-around-frontier-ai/Sat, 13 Jun 2026 12:00:00 GMTInfrastructureWhen AI Becomes Strategic Infrastructure: A Developer's and Defense Perspectivehttps://salt19.com/ai-dispatch/stories/evomind-a-governed-cognitive-architecture-for-autonomous-reasoning-planning-memory-and-desktop-a-20580153/https://salt19.com/ai-dispatch/stories/evomind-a-governed-cognitive-architecture-for-autonomous-reasoning-planning-memory-and-desktop-a-20580153/Sat, 06 Jun 2026 12:00:00 GMTCognitive SystemsEvoMind is a governed cognitive architecture and research platform for autonomous reasoning, memory, planning, learning, and desktop-agent execution. The architecture is organized around a closed-loop cognitive cycle consisting of perception, world-model maintenance, memory formation and retrieval, decision support, planning, governance enforcement,…https://salt19.com/ai-dispatch/stories/2026-01-17-ai-llms-agi-and-why-governance-matters-more-than-power/https://salt19.com/ai-dispatch/stories/2026-01-17-ai-llms-agi-and-why-governance-matters-more-than-power/Sat, 17 Jan 2026 12:00:00 GMTAGI & ArchitectureA layered view of AI, LLMs, agents, and general intelligence that argues capability and authority should be designed as separate concerns.