{"assertions":[{"assertion_index":0,"assertion_text":"Cursor Blog Cursor Blog A smart buyer takes that as a rising tide and sizes next year's seat budget to it.","citation_check":null,"claim_ids":[33376,62634,62642],"link_only_urls":[]},{"assertion_index":1,"assertion_text":"Cursor Blog Nate Jones A mean drawn from that distribution is a statement about a handful of people.","citation_check":null,"claim_ids":[27220,68432,68433,68434],"link_only_urls":[]},{"assertion_index":2,"assertion_text":"Nate Jones Whether those changes are the kind your team needs, the evidence cannot show.","citation_check":null,"claim_ids":[68432,68433,68434],"link_only_urls":[]},{"assertion_index":3,"assertion_text":"Cursor Blog Cursor Blog AI Engineer Those are different curves, and a frontier shop's case study says almost nothing about your team's shape.","citation_check":null,"claim_ids":[20046,39738,55620],"link_only_urls":[]},{"assertion_index":4,"assertion_text":"AI Engineer That is the constraint starting to bind, not a forecast of it.","citation_check":null,"claim_ids":[37228],"link_only_urls":[]},{"assertion_index":5,"assertion_text":"Don't Worry About the Vase Anthropic pays for that work directly, and the two firms have each put a floor of $1 billion, spread across five years, behind building out evaluation capacity.","citation_check":null,"claim_ids":[68460,68461,68462,68463,68464],"link_only_urls":[]},{"assertion_index":6,"assertion_text":"Anthropic News So far, then, the lab picked and pays its own evaluator.","citation_check":null,"claim_ids":[63612,63614,63615,63617],"link_only_urls":[]},{"assertion_index":7,"assertion_text":"Wiz Research This is not a new pattern.","citation_check":null,"claim_ids":[68405,68406,68407,68414,68415,68423,68426],"link_only_urls":[]},{"assertion_index":8,"assertion_text":"Microsoft Security Blog The stake for an engineer building agents is direct: an agent holding long-lived keys on a laptop is the target, and the decision is short-lived credentials before any new tool integration.","citation_check":null,"claim_ids":[42586,42595,42599],"link_only_urls":[]},{"assertion_index":9,"assertion_text":"Latent Space OpenRouter Blog The kept promise matters less than the label on the next model.","citation_check":null,"claim_ids":[41790,41791,41792,68482,68483,68484,68503,68504],"link_only_urls":[]},{"assertion_index":10,"assertion_text":"OpenRouter Blog Run 3.8 now, because the weights are real and downloadable.","citation_check":null,"claim_ids":[68482,68483,68484,68503,68504],"link_only_urls":[]},{"assertion_index":11,"assertion_text":"- Anthropic said on September 18 that more evaluator partnerships would follow within weeks, and the names will show whether any evaluator is neither a paid vendor nor a customer. Anthropic News","citation_check":null,"claim_ids":[63612,63614,63615,63617],"link_only_urls":[]},{"assertion_index":12,"assertion_text":"- Dario Amodei's scheduled White House dinner with President Trump lands in the same week the evaluator commitments firmed up; any readout on oversight is the thing to watch. TechCrunch AI","citation_check":null,"claim_ids":[68449],"link_only_urls":[]},{"assertion_index":13,"assertion_text":"- Hugging Face's Transformers library added Qwen4-Exp support in August, so the model code exists before the weights do. Transformers Releases","citation_check":null,"claim_ids":[52139],"link_only_urls":[]},{"assertion_index":14,"assertion_text":"- Sakana AI's PC-ALM trains 1,000-layer networks without backpropagation (the standard way neural networks learn), and Sakana says it may prove useful on neuromorphic chips (hardware built to mimic how brain neurons signal), where running these dynamics is cheaper than on GPUs; the test is a result at a size someone ships. Sakana AI","citation_check":null,"claim_ids":[68210,68211,68212,68213,68214,68215,68216,68217,68218],"link_only_urls":[]},{"assertion_index":15,"assertion_text":"- Reply bots that plagued Twitter have surfaced on Bluesky, and a reply bot checker Simon Willison flagged uses Bluesky's freely available API (the open interface other programs can query) to spot them by reply speed and absence of original posts. Simon Willison's Weblog","citation_check":null,"claim_ids":[68440,68441,68443],"link_only_urls":[]},{"assertion_index":16,"assertion_text":"Nate Jones The review-quality decline comes from a single conference talk without a stated dataset.","citation_check":null,"claim_ids":[68432,68433,68434],"link_only_urls":[]}],"brief_date":"2026-09-28","bundle_id":"evidence-bundle-v2-142fd388e2d6feb8d31ef149a20b049746be9c5201cdf5551bac3107a9ca4cfa","claims":[{"assertion_indexes":[0],"citation_authority_claim_id":33376,"citation_checks":[],"claim_id":33376,"claim_text":"The average Coinbase engineer has increased the number of pull requests merged by 55% since the beginning of 2026.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-07-22T03:00:05.168945+00:00","source_name":"Cursor Blog","source_url":"https://cursor.com/blog/coinbase"},{"assertion_indexes":[0],"citation_authority_claim_id":62634,"citation_checks":[],"claim_id":62634,"claim_text":"Approximately 98% of employees in Grab's tech organization used the AI coding tool Cursor monthly, and about 75% used it weekly as of the report's publication.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-18T06:33:15.211034+00:00","source_name":"Cursor Blog","source_url":"https://cursor.com/blog/grab"},{"assertion_indexes":[0],"citation_authority_claim_id":62642,"citation_checks":[],"claim_id":62642,"claim_text":"Over a third of merge requests at Grab incorporate the use of Cursor.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-18T06:33:15.211034+00:00","source_name":"Cursor Blog","source_url":"https://cursor.com/blog/grab"},{"assertion_indexes":[1],"citation_authority_claim_id":27220,"citation_checks":[],"claim_id":27220,"claim_text":"A Developer Habits Report found that top-tier developers (p99) produced 46 times more AI-assisted lines per day and merged 15 times more pull requests per week than the median active user.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-07-18T16:32:54.754674+00:00","source_name":"Cursor Blog","source_url":"https://cursor.com/blog/cfo-council"},{"assertion_indexes":[1,2,16],"citation_authority_claim_id":68432,"citation_checks":[],"claim_id":68432,"claim_text":"Cursor's Spring 2026 developer-habits report found that the 99th-percentile user merged about fifteen times as many pull requests as the median active pull-request author.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:45:23.175997+00:00","source_name":"Nate Jones","source_url":"https://natesnewsletter.substack.com/p/scale-ai-developer-productivity"},{"assertion_indexes":[1,2,16],"citation_authority_claim_id":68433,"citation_checks":[],"claim_id":68433,"claim_text":"Lauren Tan reported producing 1,000 merged pull requests in July 2026 and being on track to double that figure.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:45:23.175997+00:00","source_name":"Nate Jones","source_url":"https://natesnewsletter.substack.com/p/scale-ai-developer-productivity"},{"assertion_indexes":[1,2,16],"citation_authority_claim_id":68434,"citation_checks":[],"claim_id":68434,"claim_text":"Lauren Tan reported 2,462 merged pull requests in August 2026.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:45:23.175997+00:00","source_name":"Nate Jones","source_url":"https://natesnewsletter.substack.com/p/scale-ai-developer-productivity"},{"assertion_indexes":[3],"citation_authority_claim_id":20046,"citation_checks":[],"claim_id":20046,"claim_text":"Autonomous coding agents accounted for less than 2% of merged pull requests over the analyzed period.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-07-11T04:30:29.031000+00:00","source_name":"AI Engineer","source_url":"https://www.youtube.com/watch?v=WqZq8L-v9pA"},{"assertion_indexes":[3],"citation_authority_claim_id":39738,"citation_checks":[],"claim_id":39738,"claim_text":"In December, cloud agents authored approximately 10% of the pull requests merged to the Cursor monorepo. (Source: Cursor blog post, published 2026-07-30)","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-07-31T06:23:00.797674+00:00","source_name":"Cursor Blog","source_url":"https://cursor.com/blog/cloud-agent-environment"},{"assertion_indexes":[3],"citation_authority_claim_id":55620,"citation_checks":[],"claim_id":55620,"claim_text":"Cursor reports that cloud agents create more than 60% of the pull requests merged internally by the company.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-03T06:40:40.824962+00:00","source_name":"Cursor Blog","source_url":"https://cursor.com/blog/self-hosted-machines"},{"assertion_indexes":[4],"citation_authority_claim_id":37228,"citation_checks":[],"claim_id":37228,"claim_text":"Since the widespread adoption of AI coding tools in early 2026, pull request code review quality has declined, characterized by increased comment length and a rise in unreviewed merges.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-07-26T06:23:34.200917+00:00","source_name":"AI Engineer","source_url":"https://www.youtube.com/watch?v=Ib5GBkD555M"},{"assertion_indexes":[5],"citation_authority_claim_id":68460,"citation_checks":[],"claim_id":68460,"claim_text":"Dario Amodei committed Anthropic to using embedded evaluators with employee-level access to provide outside perspective and reports.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:46:45.460514+00:00","source_name":"Don't Worry About the Vase","source_url":"https://thezvi.substack.com/p/the-quest-for-embedded-evaluators"},{"assertion_indexes":[5],"citation_authority_claim_id":68461,"citation_checks":[],"claim_id":68461,"claim_text":"OpenAI committed to embedded evaluators and issued a call for international coordination.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:46:45.460514+00:00","source_name":"Don't Worry About the Vase","source_url":"https://thezvi.substack.com/p/the-quest-for-embedded-evaluators"},{"assertion_indexes":[5],"citation_authority_claim_id":68462,"citation_checks":[],"claim_id":68462,"claim_text":"Anthropic announced a partnership with Accenture's Faculty subsidiary for embedded evaluation, including evaluating, red-teaming, and alignment assessments.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:46:45.460514+00:00","source_name":"Don't Worry About the Vase","source_url":"https://thezvi.substack.com/p/the-quest-for-embedded-evaluators"},{"assertion_indexes":[5],"citation_authority_claim_id":68463,"citation_checks":[],"claim_id":68463,"claim_text":"Anthropic and Accenture each expect to invest at least $1 billion in building capacity for embedded evaluation over the next five years.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:46:45.460514+00:00","source_name":"Don't Worry About the Vase","source_url":"https://thezvi.substack.com/p/the-quest-for-embedded-evaluators"},{"assertion_indexes":[5],"citation_authority_claim_id":68464,"citation_checks":[],"claim_id":68464,"claim_text":"Anthropic will directly fund Accenture's work on embedded evaluation.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:46:45.460514+00:00","source_name":"Don't Worry About the Vase","source_url":"https://thezvi.substack.com/p/the-quest-for-embedded-evaluators"},{"assertion_indexes":[6,11],"citation_authority_claim_id":63612,"citation_checks":[],"claim_id":63612,"claim_text":"Anthropic has entered into a partnership with Accenture to conduct independent embedded evaluation of its frontier AI models.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-19T06:46:09.827182+00:00","source_name":"Anthropic News","source_url":"https://www.anthropic.com/news/accenture-embedded-evaluation"},{"assertion_indexes":[6,11],"citation_authority_claim_id":63614,"citation_checks":[],"claim_id":63614,"claim_text":"Anthropic and Accenture each intend to invest at least $1 billion in building capacity for embedded evaluation over the next five years.","correction_role":null,"corroborates_eligible":true,"label":"forecast","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-19T06:46:09.827182+00:00","source_name":"Anthropic News","source_url":"https://www.anthropic.com/news/accenture-embedded-evaluation"},{"assertion_indexes":[6,11],"citation_authority_claim_id":63615,"citation_checks":[],"claim_id":63615,"claim_text":"Anthropic plans to fund Accenture's embedded evaluation work directly.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-19T06:46:09.827182+00:00","source_name":"Anthropic News","source_url":"https://www.anthropic.com/news/accenture-embedded-evaluation"},{"assertion_indexes":[6,11],"citation_authority_claim_id":63617,"citation_checks":[],"claim_id":63617,"claim_text":"Anthropic expects to announce additional evaluator partnerships in the coming weeks.","correction_role":null,"corroborates_eligible":true,"label":"forecast","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-19T06:46:09.827182+00:00","source_name":"Anthropic News","source_url":"https://www.anthropic.com/news/accenture-embedded-evaluation"},{"assertion_indexes":[7],"citation_authority_claim_id":68405,"citation_checks":[],"claim_id":68405,"claim_text":"Infostealer malware is a leading initial access vector for compromising enterprise cloud, code, and AI environments.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:44:36.822666+00:00","source_name":"Wiz Research","source_url":"https://www.wiz.io/blog/infostealer-incursion-cloud-ai-credentials"},{"assertion_indexes":[7],"citation_authority_claim_id":68406,"citation_checks":[],"claim_id":68406,"claim_text":"Microsoft, Recorded Future, and Verizon\u2019s DBIR identify infostealers as a major cause of concern regarding initial infection vectors.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:44:36.822666+00:00","source_name":"Wiz Research","source_url":"https://www.wiz.io/blog/infostealer-incursion-cloud-ai-credentials"},{"assertion_indexes":[7],"citation_authority_claim_id":68407,"citation_checks":[],"claim_id":68407,"claim_text":"Lumma C2, RedLine, and Vidar infostealer families account for 85.7% of detected incidents in Wiz\u2019s analysis of NordStellar data.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:44:36.822666+00:00","source_name":"Wiz Research","source_url":"https://www.wiz.io/blog/infostealer-incursion-cloud-ai-credentials"},{"assertion_indexes":[7],"citation_authority_claim_id":68414,"citation_checks":[],"claim_id":68414,"claim_text":"Infostealer-like malware such as Miasma has recently targeted CI/CD pipelines, development servers, and build processes.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:44:36.822666+00:00","source_name":"Wiz Research","source_url":"https://www.wiz.io/blog/infostealer-incursion-cloud-ai-credentials"},{"assertion_indexes":[7],"citation_authority_claim_id":68415,"citation_checks":[],"claim_id":68415,"claim_text":"Infostealer malware steals active session tokens to bypass multi-factor authentication by hijacking already-authenticated sessions.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:44:36.822666+00:00","source_name":"Wiz Research","source_url":"https://www.wiz.io/blog/infostealer-incursion-cloud-ai-credentials"},{"assertion_indexes":[7],"citation_authority_claim_id":68423,"citation_checks":[],"claim_id":68423,"claim_text":"Long-lived, portable service account key files are a top target for infostealers scanning developer machines in GCP environments.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:44:36.822666+00:00","source_name":"Wiz Research","source_url":"https://www.wiz.io/blog/infostealer-incursion-cloud-ai-credentials"},{"assertion_indexes":[7],"citation_authority_claim_id":68426,"citation_checks":[],"claim_id":68426,"claim_text":"Around 50% of infostealer malware families now actively collect GitHub credentials.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:44:36.822666+00:00","source_name":"Wiz Research","source_url":"https://www.wiz.io/blog/infostealer-incursion-cloud-ai-credentials"},{"assertion_indexes":[8],"citation_authority_claim_id":42586,"citation_checks":[],"claim_id":42586,"claim_text":"The malware searches developer workstations and CI/CD environments for npm, GitHub, cloud, and infrastructure credentials.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-06T06:22:01.138762+00:00","source_name":"Microsoft Security Blog","source_url":"https://www.microsoft.com/en-us/security/blog/2026/08/04/chaindrop-supply-chain-compromise-anatomy-self-propagating-worm/"},{"assertion_indexes":[8],"citation_authority_claim_id":42595,"citation_checks":[],"claim_id":42595,"claim_text":"Evidence points towards stolen maintainer credentials as the attack vector for initial compromise.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-06T06:22:01.138762+00:00","source_name":"Microsoft Security Blog","source_url":"https://www.microsoft.com/en-us/security/blog/2026/08/04/chaindrop-supply-chain-compromise-anatomy-self-propagating-worm/"},{"assertion_indexes":[8],"citation_authority_claim_id":42599,"citation_checks":[],"claim_id":42599,"claim_text":"The malware's cloud and secret store enumeration modules use available credentials to call service APIs, verify access, and retrieve additional secrets permitted to those identities.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-06T06:22:01.138762+00:00","source_name":"Microsoft Security Blog","source_url":"https://www.microsoft.com/en-us/security/blog/2026/08/04/chaindrop-supply-chain-compromise-anatomy-self-propagating-worm/"},{"assertion_indexes":[9],"citation_authority_claim_id":41790,"citation_checks":[],"claim_id":41790,"claim_text":"Alibaba announced Qwen3.8-Max, a 2.4T-parameter model, and stated that open weights for this model would be released the week following the announcement.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-04T06:58:57.201174+00:00","source_name":"Latent Space","source_url":"https://www.latent.space/p/ainews-qwen-38-max24t-and-27b-new"},{"assertion_indexes":[9],"citation_authority_claim_id":41791,"citation_checks":[],"claim_id":41791,"claim_text":"Alibaba announced that the Qwen3.8-27B model would also be released with open weights.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-04T06:58:57.201174+00:00","source_name":"Latent Space","source_url":"https://www.latent.space/p/ainews-qwen-38-max24t-and-27b-new"},{"assertion_indexes":[9],"citation_authority_claim_id":41792,"citation_checks":[],"claim_id":41792,"claim_text":"Alibaba disclosed API pricing for Qwen3.8-Max as $2.00 per million input tokens, $6.00 per million output tokens, and $0.25 per million cached tokens.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-04T06:58:57.201174+00:00","source_name":"Latent Space","source_url":"https://www.latent.space/p/ainews-qwen-38-max24t-and-27b-new"},{"assertion_indexes":[9,10],"citation_authority_claim_id":68482,"citation_checks":[],"claim_id":68482,"claim_text":"Alibaba released the Qwen 3.8 model generation in August 2026.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:48:05.999731+00:00","source_name":"OpenRouter Blog","source_url":"https://openrouter.ai/blog/insights/qwen-3-8/"},{"assertion_indexes":[9,10],"citation_authority_claim_id":68483,"citation_checks":[],"claim_id":68483,"claim_text":"The Qwen3.8-2.4T-A95B model weights were published to Hugging Face on August 12, 2026.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:48:05.999731+00:00","source_name":"OpenRouter Blog","source_url":"https://openrouter.ai/blog/insights/qwen-3-8/"},{"assertion_indexes":[9,10],"citation_authority_claim_id":68484,"citation_checks":[],"claim_id":68484,"claim_text":"The Qwen3.8-27B model weights were published to Hugging Face on August 14, 2026.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:48:05.999731+00:00","source_name":"OpenRouter Blog","source_url":"https://openrouter.ai/blog/insights/qwen-3-8/"},{"assertion_indexes":[9,10],"citation_authority_claim_id":68503,"citation_checks":[],"claim_id":68503,"claim_text":"Qwen 3.8 follows Qwen 3.7, with Qwen 3.5, 3.6, and 3.7 released between the Qwen 3 generation (April 2025) and Qwen 3.8.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:48:05.999731+00:00","source_name":"OpenRouter Blog","source_url":"https://openrouter.ai/blog/insights/qwen-3-8/"},{"assertion_indexes":[9,10],"citation_authority_claim_id":68504,"citation_checks":[],"claim_id":68504,"claim_text":"The Qwen3.8-Flash-Next model is described by Alibaba as an experimental preview of the architecture that will underpin Qwen 4.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:48:05.999731+00:00","source_name":"OpenRouter Blog","source_url":"https://openrouter.ai/blog/insights/qwen-3-8/"},{"assertion_indexes":[12],"citation_authority_claim_id":68449,"citation_checks":[],"claim_id":68449,"claim_text":"Anthropic CEO Dario Amodei is scheduled to have dinner with President Donald Trump at the White House.","correction_role":null,"corroborates_eligible":true,"label":"forecast","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:46:23.431455+00:00","source_name":"TechCrunch AI","source_url":"https://techcrunch.com/2026/09/27/anthropics-ceo-is-about-to-have-dinner-with-president-trump/"},{"assertion_indexes":[13],"citation_authority_claim_id":52139,"citation_checks":[],"claim_id":52139,"claim_text":"The Hugging Face Transformers library version 5.16.0, published on 2026-08-26, added support for the Qwen4-Exp model.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-27T06:42:28.697351+00:00","source_name":"Transformers Releases","source_url":"https://github.com/huggingface/transformers/releases/tag/v5.16.0"},{"assertion_indexes":[14],"citation_authority_claim_id":68210,"citation_checks":[],"claim_id":68210,"claim_text":"Sakana AI introduced PC-ALM, a method for training neural networks that serves as an alternative to backpropagation.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68211,"citation_checks":[],"claim_id":68211,"claim_text":"The PC-ALM method is capable of training neural networks with 1000 layers.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68212,"citation_checks":[],"claim_id":68212,"claim_text":"PC-ALM trains neural networks using only local dynamics and without backpropagation.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68213,"citation_checks":[],"claim_id":68213,"claim_text":"PC-ALM introduces dual neurons, which are Lagrange multipliers, as part of the layer-local dynamics.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68214,"citation_checks":[],"claim_id":68214,"claim_text":"In PC-ALM, each layer acts as a PI feedback control system to minimize local prediction errors.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68215,"citation_checks":[],"claim_id":68215,"claim_text":"PC-ALM is capable of propagating signals to seemingly arbitrary depth.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68216,"citation_checks":[],"claim_id":68216,"claim_text":"PC-ALM is especially effective in deep narrow networks where standard predictive coding struggles to learn.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68217,"citation_checks":[],"claim_id":68217,"claim_text":"The motivation for PC-ALM is to understand how distributed physical systems, such as the brain, can compute credit signals using only local coupling and local dynamics.","correction_role":null,"corroborates_eligible":true,"label":"opinion","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[14],"citation_authority_claim_id":68218,"citation_checks":[],"claim_id":68218,"claim_text":"PC-ALM may inform deep learning in neuromorphic hardware, where dynamics are cheaper than on GPUs.","correction_role":null,"corroborates_eligible":true,"label":"forecast","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:36:45.025800+00:00","source_name":"Sakana AI","source_url":"https://sakana.ai/pc-alm/"},{"assertion_indexes":[15],"citation_authority_claim_id":68440,"citation_checks":[],"claim_id":68440,"claim_text":"Automated reply bots have begun appearing on the Bluesky platform, mirroring their presence on Twitter.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:45:51.275044+00:00","source_name":"Simon Willison's Weblog","source_url":"https://simonwillison.net/2026/Sep/27/bluesky-bot-check/"},{"assertion_indexes":[15],"citation_authority_claim_id":68441,"citation_checks":[],"claim_id":68441,"claim_text":"Unlike Twitter, the Bluesky platform provides a freely available and useful API.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:45:51.275044+00:00","source_name":"Simon Willison's Weblog","source_url":"https://simonwillison.net/2026/Sep/27/bluesky-bot-check/"},{"assertion_indexes":[15],"citation_authority_claim_id":68443,"citation_checks":[],"claim_id":68443,"claim_text":"The Bluesky reply bot checker identifies potential bots by analyzing signals such as replies posted within seconds of other posts from the same account, accounts that never post original content, and the use of question marks.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-28T06:45:51.275044+00:00","source_name":"Simon Willison's Weblog","source_url":"https://simonwillison.net/2026/Sep/27/bluesky-bot-check/"}],"version":"evidence-bundle-v2"}