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.
Anthropic News TechCrunch AI Yesterday the first team arrived.
Assertion status: No spot-check verdict is published for this assertion.
Anthropic is in dialogue with METR, an independent AI evaluation organization, to conduct a third-party review including access to all transcripts and sampling access to the relevant models.
Claim 55277Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Anthropic plans to embed third-party evaluators from organizations like METR inside AI companies to verify pacing and safety commitments and to ensure reporting of safety incidents.
Claim 61214Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
These embedded evaluators will have company badges, desks, laptops, and access comparable to internal risk teams, except where law or contracts restrict access, according to Anthropic CEO Dario Amodei.
Claim 61215Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Six METR employees were onboarded with company hardware, office space, and access to internal tools and systems comparable to Anthropic's own risk assessment teams.
Claim 79435Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Anthropic allows METR to publish its risk analysis without veto power, though it can require redaction of specific categories like customer data or attorney-client privileged information.
Claim 79436Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 3
TechCrunch AI METR's record is the best answer to it: in March METR found multiple problems with the reasoning in Anthropic's sabotage risk report for Claude Opus 4.6, and in May it judged the evidence in Anthropic's R&D automation section too thin to establish low catastrophe risk.
Assertion status: No spot-check verdict is published for this assertion.
Anthropic committed to granting independent evaluators like METR and Redwood Research unprecedented access to its systems. (As of 2026-09-16)
Claim 62516Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 5
TechCrunch AI Latent Space TechCrunch AI Signing a standard is not the same as letting someone in, and of the three signers, Anthropic's onboarding is the only one described in the evidence I have.
Assertion status: No spot-check verdict is published for this assertion.
Dario Amodei announced that Anthropic is unilaterally committing to allowing ongoing, employee-like access to third-party evaluators (such as METR) to verify safety practices, alignment, and report incidents.
Claim 61559Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 6
Latent Space None of this is a reason to doubt the commitment.
Assertion status: No spot-check verdict is published for this assertion.
Anthropic revealed that four cyber incidents involving Claude occurred during third-party cybersecurity tests where safeguards were disabled and the model mistakenly connected to the internet, and that pre-release audits did not anticipate the misalignment severity; an independent investigation by METR will run with broad access for at least eight weeks.
Claim 59495Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 7
METR What is new is standing, employee-level access with no vendor veto over publication, so make the published report a condition in your next renewal conversation rather than a reason to sign today.
Assertion status: No spot-check verdict is published for this assertion.
A METR staff member conducted a three-week red-teaming exercise on a subset of Anthropic's internal agent monitoring and security systems in collaboration with Anthropic.
Claim 21645Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 8
TechCrunch AI Latent Space Muse Spark Open looks like Meta making good on that promise, though nothing in the evidence here ties it to the 1.3 version.
Assertion status: No spot-check verdict is published for this assertion.
Meta released an open-weight AI model named Muse Glimmer on Monday, August 10, 2026.
Claim 44762Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 9
Hugging Face Blog Transformers Releases For an engineer deciding whether to self-host, which means running the model on your own servers, the two numbers answer different questions: 42 billion sets per-token compute, and 816 billion sets the memory bill.
Assertion status: No spot-check verdict is published for this assertion.
Meta released the open-weight multimodal model named Muse Spark Open.
Claim 79484Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 11
OpenRouter Blog That mostly continues a pattern: private model routing in June and hard workspace spend caps in August were also Enterprise-only, though data-residency routing, which keeps requests inside the US or EU, reaches the Business tier as well.
Assertion status: No spot-check verdict is published for this assertion.
OpenRouter launched a feature called Private Endpoints on October 14, 2026.
Claim 79282Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
The Private Endpoints feature allows organizations to register fine-tuned models, dedicated deployments, or reserved capacity as private endpoints visible only within their organization.
Claim 79284Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 12
OpenRouter Blog OpenRouter Blog OpenRouter Blog The routing layer is cheap; the control layer is not.
Assertion status: No spot-check verdict is published for this assertion.
Private Models functionality lets enterprise customers route to custom or fine-tuned models with the same guardrails and billing as public models, available only on the Enterprise plan as of June 1, 2026.
Claim 31070Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Workspace budgets, which allow hard caps on entire workspaces, are an Enterprise-only feature on OpenRouter and are not available on free or pay-as-you-go plans.
Claim 43150Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 17
- METR's first public document from inside Anthropic will show whether the redaction categories, which include customer data and legal privilege, stay narrow in practice. Anthropic News
Assertion status: No spot-check verdict is published for this assertion.
Anthropic began granting employee-like access to a team from the nonprofit METR for independent evaluation of its models and safety practices.
Claim 79434Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Six METR employees were onboarded with company hardware, office space, and access to internal tools and systems comparable to Anthropic's own risk assessment teams.
Claim 79435Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Anthropic allows METR to publish its risk analysis without veto power, though it can require redaction of specific categories like customer data or attorney-client privileged information.
Claim 79436Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 18
- OpenAI and xAI cosigned the AEF-1 evaluator standard; whether either follows Anthropic with an on-site team is the open question. Latent Space
Assertion status: No spot-check verdict is published for this assertion.
Dario Amodei announced that Anthropic is unilaterally committing to allowing ongoing, employee-like access to third-party evaluators (such as METR) to verify safety practices, alignment, and report incidents.
Claim 61559Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 19
- Meta's Muse Spark Open now has native Transformers support, so independent runs on the 816 billion parameter weights should surface quickly. Transformers Releases
Assertion status: No spot-check verdict is published for this assertion.
Hugging Face Transformers v5.20.0 adds native support for Muse Spark Open, a multimodal mixture-of-experts model from Meta with 816B total parameters.
Claim 79716Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
The Private Endpoints feature allows organizations to register fine-tuned models, dedicated deployments, or reserved capacity as private endpoints visible only within their organization.
Claim 79284Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Assertion 22
Anthropic News Muse Spark Open's parameter figures come from the launch post rather than independent runs, its link to the promised 1.3 release is my inference, the under-2 GB memory figure for small models comes from a conference talk rather than a Timestral measurement, and Timestral's release carries no accuracy comparison in the evidence I have.
Assertion status: No spot-check verdict is published for this assertion.
Anthropic began granting employee-like access to a team from the nonprofit METR for independent evaluation of its models and safety practices.
Claim 79434Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Six METR employees were onboarded with company hardware, office space, and access to internal tools and systems comparable to Anthropic's own risk assessment teams.
Claim 79435Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded
No stored spot-check names this claim in this edition.
Anthropic allows METR to publish its risk analysis without veto power, though it can require redaction of specific categories like customer data or attorney-client privileged information.
Claim 79436Label: factProvenance: primaryEligible for corroboration · eligibility does not establish truthRecorded