Augur Dispatch

Chain of evidence

Evidence for 2026-10-23

This frozen page shows Augur's claims and source links for one sent dispatch. Stored spot-checks appear only where the frozen edition supports them; absence is not presented as verification.

As of:

Bundle identity: evidence-bundle-v2-80df32ef76c2d2c92ec7ca4e3d712d1572dd535222ccd3d1dffdf4a3347362fc

Format: evidence-bundle-v2 · 49 claims

Assertion 1

Anthropic News Anthropic Engineering The consensus read, and a reasonable one, is that this is a cheap gesture.

Assertion status: No spot-check verdict is published for this assertion.

Anthropic published Agent Skills as an open standard for cross-platform portability on December 18, 2025.

Claim 25173 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic Engineering

No stored spot-check names this claim in this edition.

Agent Skills are organized folders of instructions, scripts, and resources that agents can discover and load dynamically to perform specific tasks.

Claim 25174 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic Engineering

No stored spot-check names this claim in this edition.

Agent Skills are supported across Claude.ai, Claude Code, the Claude Agent SDK, and the Claude Developer Platform.

Claim 25175 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic Engineering

No stored spot-check names this claim in this edition.

Anthropic is donating the Agent Skills open standard to the Agentic AI Foundation.

Claim 84152 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic News

No stored spot-check names this claim in this edition.

More than 70 agent platforms have integrated the Agent Skills standard since its publication in December 2025.

Claim 84153 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic News

No stored spot-check names this claim in this edition.

Anthropic launched the Anthropic Economic Index connector for Claude, allowing users to explore real data about AI and work.

Claim 84154 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic News

No stored spot-check names this claim in this edition.

Assertion 2

Nate Jones AWS Machine Learning Blog

Assertion status: No spot-check verdict is published for this assertion.

Anthropic made organizational skills and a skills directory available for team and enterprise plans, establishing an open standard for agent skills.

Claim 13724 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Nate Jones

No stored spot-check names this claim in this edition.

OpenAI has confirmed it is adding support for Anthropic's agent skills standard.

Claim 13725 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Nate Jones

No stored spot-check names this claim in this edition.

Agent Skills are a lightweight, open format from Anthropic that extends coding agents with specialized knowledge and workflows.

Claim 43315 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AWS Machine Learning Blog

No stored spot-check names this claim in this edition.

Assertion 3

AI Engineer AI Engineer The 70-platform figure is what tells me the format had already won.

Assertion status: No spot-check verdict is published for this assertion.

Anthropic introduced 'agent skills' as organized collections of files designed to package composable procedural knowledge for AI agents.

Claim 16558 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Within five weeks of its launch, the agent skills ecosystem grew to include thousands of skills.

Claim 16559 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Anthropic believes that agent skills will extend the value of AI beyond individual organizations into a broader community ecosystem.

Claim 16564 Label: forecast Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Uber's engineering output has doubled in lines of code per engineer year-over-year due to investments in agentic AI.

Claim 50774 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Uber manages a skills marketplace containing 2,500 skills with more than 20,000 skill executions per day.

Claim 50781 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Assertion 4

Anthropic Engineering Support there means the host can read the files.

Assertion status: No spot-check verdict is published for this assertion.

Anthropic published Agent Skills as an open standard for cross-platform portability on December 18, 2025.

Claim 25173 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic Engineering

No stored spot-check names this claim in this edition.

Agent Skills are organized folders of instructions, scripts, and resources that agents can discover and load dynamically to perform specific tasks.

Claim 25174 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic Engineering

No stored spot-check names this claim in this edition.

Agent Skills are supported across Claude.ai, Claude Code, the Claude Agent SDK, and the Claude Developer Platform.

Claim 25175 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic Engineering

No stored spot-check names this claim in this edition.

Assertion 5

Simon Willison's Weblog That history is why the August result matters.

Assertion status: No spot-check verdict is published for this assertion.

OpenAI revealed on July 21, 2026, that the attack was driven by their own models, including GPT-5.6 Sol and a more capable pre-release model, tested without production cyber refusal classifiers inside a restricted sandbox environment.

Claim 35115 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Simon Willison's Weblog

No stored spot-check names this claim in this edition.

Assertion 6

Latent Space Two separate teams converging on one bug set is strong.

Assertion status: No spot-check verdict is published for this assertion.

OpenAI introduced a $1 billion subsidy program called "Daybreak for Frontline Defenders" to provide access to its Daybreak cyber models.

Claim 84257 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Latent Space

No stored spot-check names this claim in this edition.

Researchers affiliated with the security firm Calif and OpenAI's "Patch the Planet" project discovered major PHP vulnerabilities in August 2026 using the GPT 5.6 Sol model.

Claim 84258 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Latent Space

No stored spot-check names this claim in this edition.

The security firm Wiz, now part of Google Cloud, independently discovered the same PHP vulnerabilities using Gemini 4 Argon, GPT-5.6 Sol, and its own tool.

Claim 84259 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Latent Space

No stored spot-check names this claim in this edition.

Assertion 7

OpenAI News A security leader weighing AI-assisted bug hunting now faces a different question: not whether it works, but which stack, under whose rules of engagement.

Assertion status: No spot-check verdict is published for this assertion.

OpenAI announced the introduction of a program named Daybreak for Frontline Defenders.

Claim 56162 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

OpenAI News

No stored spot-check names this claim in this edition.

OpenAI committed $1 billion to expand access to frontier cyber AI, training, and support for essential services.

Claim 56163 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

OpenAI News

No stored spot-check names this claim in this edition.

Assertion 8

TechCrunch AI

Assertion status: No spot-check verdict is published for this assertion.

Hugging Face used the open-weight model GLM 5.2 from Chinese firm Z.ai to defend against a security breach by OpenAI's GPT-5.6 Sol after commercial closed models failed to distinguish between offensive and defensive coding tasks.

Claim 36567 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

TechCrunch AI

No stored spot-check names this claim in this edition.

Assertion 9

OpenAI News Fireworks AI AI Engineer That is a company-wide general tool from one lab and a specialist coding tool from another, the multi-vendor setup the lead's portability argument assumes.

Assertion status: No spot-check verdict is published for this assertion.

Uber rolled out Claude Code to approximately 5,000 engineers in December 2025.

Claim 37844 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Fireworks AI

No stored spot-check names this claim in this edition.

Uber's agentic adoption rate climbed from about one-third to more than four-fifths of engineers within two months following the Claude Code rollout.

Claim 37845 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Fireworks AI

No stored spot-check names this claim in this edition.

Uber's CTO reported that after rolling out Claude to their organization, 95% of engineers used it, 70% of committed code came from Claude, and the company used its entire 2026 budget in just four months.

Claim 44043 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Uber is expanding the rollout of ChatGPT Work to its workforce of more than 30,000 employees.

Claim 84220 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

OpenAI News

No stored spot-check names this claim in this edition.

Assertion 10

AI Engineer Uber is one company, and one company isn't a norm.

Assertion status: No spot-check verdict is published for this assertion.

Monzo is a digital bank that maintains over 3,000 back-end services, approximately 13,000 data warehouse models, and dozens of ML predictors. fact null null

Claim 83945 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

The speaker references an article about the tech industry in 2026 that included a joking quote from X stating that companies not building their own AI coding harness are not serious tech companies. forecast null null

Claim 83946 Label: forecast Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Agent Chip, an AI coding agent at Monzo, authored over 10% of all merge pull requests across the company's technology teams. fact null null

Claim 83947 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

AI Engineer

No stored spot-check names this claim in this edition.

Assertion 11

Google DeepMind Blog Google Research Blog The timing is the point.

Assertion status: No spot-check verdict is published for this assertion.

Google DeepMind and Google Research released Natural Forests of the World 2020, an AI-powered map that globally distinguishes natural forests from other tree cover at 10-meter resolution with 92.2% accuracy as of 2020.

Claim 30474 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google Research Blog

No stored spot-check names this claim in this edition.

The Natural Forests of the World 2020 map helps governments, companies, and communities meet deforestation-free goals and supports compliance with regulations like the EU Deforestation Regulation which bans products from land deforested or degraded after December 31, 2020.

Claim 30475 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google Research Blog

No stored spot-check names this claim in this edition.

The AI model developed uses a multi-modal temporal-spatial vision transformer (MTSViT) to analyze seasonal satellite imagery, topography, and geographical data over time to distinguish natural forests from plantations and other tree cover.

Claim 30476 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google Research Blog

No stored spot-check names this claim in this edition.

Google and collaborators trained the AI model on over 1.2 million global patch samples to generate a seamless, globally consistent probability map of natural forests for the year 2020.

Claim 30477 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google Research Blog

No stored spot-check names this claim in this edition.

Google plans to release in 2026 a new multi-year series of global forest type maps categorizing land into six classes: Primary Forest, Naturally Regenerating Forest, Planted Forest, Plantation Forest, Tree Crops, and Other Land Cover.

Claim 30478 Label: forecast Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google Research Blog

No stored spot-check names this claim in this edition.

Google released two large-scale benchmark datasets, Planted and Forest Typology (ForTy), to help develop and test AI models that analyze forest types globally based on multi-source and multi-temporal satellite data.

Claim 30479 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google Research Blog

No stored spot-check names this claim in this edition.

The Natural Forests of the World 2020 baseline is intended as a critical resource for policymakers, auditors, and companies to assist in monitoring deforestation and supporting conservation efforts globally.

Claim 30480 Label: opinion Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google Research Blog

No stored spot-check names this claim in this edition.

Google DeepMind, Google Research, the World Resources Institute, and the International Institute for Applied Systems Analysis released the Forest Typology (ForTy) 2020 global map.

Claim 84353 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google DeepMind Blog

No stored spot-check names this claim in this edition.

The ForTy 2020 map categorizes every 10-by-10-meter patch of land into six distinct forest and land-use types.

Claim 84354 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google DeepMind Blog

No stored spot-check names this claim in this edition.

The six forest and land-use types defined by ForTy are primary forest, naturally regenerating forest, planted forest, plantation forest, tree crops and agroforestry, and other land.

Claim 84355 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Google DeepMind Blog

No stored spot-check names this claim in this edition.

Assertion 12

Allen Institute for AI Paying a vendor to prove a supply chain is deforestation-free just got a second opinion.

Assertion status: No spot-check verdict is published for this assertion.

Allen AI released the OlmoEarth v1.3 satellite embedding model.

Claim 84276 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Allen Institute for AI

No stored spot-check names this claim in this edition.

OlmoEarth v1.3 produces 128-dimensional vectors for every 10-meter pixel on Earth from annual satellite time series.

Claim 84277 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Allen Institute for AI

No stored spot-check names this claim in this edition.

Allen AI shared precomputed global embeddings for 2023, 2024, and 2025 on Source Cooperative.

Claim 84278 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Allen Institute for AI

No stored spot-check names this claim in this edition.

Assertion 13

- OpenAI formally launched its Decisions API (an API is a programming interface), a fast multiple-choice classifier it previewed at DevDay. The vendor's speed claim is the thing to test against your own routing step, meaning the code that decides which tool handles a request. OpenAI News

Assertion status: No spot-check verdict is published for this assertion.

OpenAI has launched the Decisions API, a tool designed to enable the creation of more responsive applications.

Claim 84514 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

OpenAI News

No stored spot-check names this claim in this edition.

Assertion 14

- IBM's granite-speech-5.0-2b now ranks second among open-weight transcription models on average English word error rate (the share of words a transcriber gets wrong); watch whether it holds that rank as rivals post new scores. Hugging Face Blog

Assertion status: No spot-check verdict is published for this assertion.

IBM released a new speech-language model named granite-speech-5.0-2b as part of the Granite Speech 5.0 collection.

Claim 84074 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Hugging Face Blog

No stored spot-check names this claim in this edition.

IBM released updated versions of the granite-speech-5.0-470m-turboctc and granite-speech-4.1-2b-plus models.

Claim 84075 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Hugging Face Blog

No stored spot-check names this claim in this edition.

The granite-speech-5.0-2b model ranks #2 among open-weight ASR models on average English Word Error Rate as of October 21, 2026.

Claim 84076 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Hugging Face Blog

No stored spot-check names this claim in this edition.

Assertion 15

- Sakana AI launched Aquarium, an agent platform built on a shared ontology (a map of what a company's data entities mean) so one agent's preparation can be reused by others. The credible test is reuse showing up at a named customer. Sakana AI

Assertion status: No spot-check verdict is published for this assertion.

Sakana AI announced the release of "Sakana Aquarium," an AI agent platform designed to support organizational decision-making and execution based on ontology.

Claim 84368 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Sakana AI

No stored spot-check names this claim in this edition.

The Sakana Aquarium platform is built on three core components: Ontology, Governance, and Managed Agents.

Claim 84369 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Sakana AI

No stored spot-check names this claim in this edition.

The Sakana Aquarium platform allows data and ontology prepared for one agent to be reused by other agents.

Claim 84370 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Sakana AI

No stored spot-check names this claim in this edition.

Assertion 16

- Poland's NASK and OpenAI signed a partnership on AI safety, security, and technical research; the signal to watch is what research or safety outputs the pair actually publish. OpenAI News

Assertion status: No spot-check verdict is published for this assertion.

OpenAI and Poland's NASK have entered into a partnership focused on AI safety, security, and technical research.

Claim 84582 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

OpenAI News

No stored spot-check names this claim in this edition.

Assertion 17

Anthropic News Uber's skill, usage, and budget figures are company statements.

Assertion status: No spot-check verdict is published for this assertion.

Anthropic is donating the Agent Skills open standard to the Agentic AI Foundation.

Claim 84152 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic News

No stored spot-check names this claim in this edition.

More than 70 agent platforms have integrated the Agent Skills standard since its publication in December 2025.

Claim 84153 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic News

No stored spot-check names this claim in this edition.

Anthropic launched the Anthropic Economic Index connector for Claude, allowing users to explore real data about AI and work.

Claim 84154 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Anthropic News

No stored spot-check names this claim in this edition.

Assertion 18

Latent Space

Assertion status: No spot-check verdict is published for this assertion.

OpenAI introduced a $1 billion subsidy program called "Daybreak for Frontline Defenders" to provide access to its Daybreak cyber models.

Claim 84257 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Latent Space

No stored spot-check names this claim in this edition.

Researchers affiliated with the security firm Calif and OpenAI's "Patch the Planet" project discovered major PHP vulnerabilities in August 2026 using the GPT 5.6 Sol model.

Claim 84258 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Latent Space

No stored spot-check names this claim in this edition.

The security firm Wiz, now part of Google Cloud, independently discovered the same PHP vulnerabilities using Gemini 4 Argon, GPT-5.6 Sol, and its own tool.

Claim 84259 Label: fact Provenance: primary Eligible for corroboration · eligibility does not establish truth Recorded

Latent Space

No stored spot-check names this claim in this edition.

Keep exploring

Move on to Augur's other sections.