AI social digest — 2026-10-02
Top line: An X-layer day with one blunt verdict: the agent race got its first three-way independent judgment (Grok Bot over Muse and Dots), OpenClaw's marketplace hardened its security layer, and the Anthropic machine-consciousness fight picked up a theology problem. Economics discourse (Rattner, Acemoglu) keeps building the ai-economics layer underneath.
1. The agent race, judged: Grok Bot > Muse > Dots (single reviewer — flag)
Morgan (~7:34 PM ET): "after three days with Dot, I have to say, I am really impressed with it, but I do think Grok Bot is a better product at this stage" — "better than muse and better than dots by a pretty clear margin in my opinion," more details promised. Related in the same window: the reviewer also found the Dots phone flow "gimmicky, somewhat slow and inefficient" (dots-demo thread, ~12:21 AM ET).
What's genuinely new: the first three-way verdict from an apparent user of all three. Full handling (with the Verge's "enterprise software that can also order your dinner" hands-on) in the ai digest. Caveats do the heavy lifting: one reviewer, X-grade, no methodology, "better" undefined — but the race now has a non-keynote datapoint, and "more details coming" is the watch.
Link: Morgan review (feed story 6455) · dots demo (story 6374)
2. OpenClaw hardens the marketplace: vendor plugins + Tencent AIG security review
Two adjacent posts (~11:22 PM ET): (1) Patrick Erichsen — 50+ plugins added to OpenClaw from Notion, Canva, Dropbox et al., installed directly from chat, hot-reloaded, "no gateway restart needed"; (2) ClawScan (the OSS CLI powering ClawHub security review) now integrates Tencent Hunyuan's AI-Infra-Guard (AIG) — every skill/plugin upload to ClawHub runs through it. Same-week context: the AI Worth Using × OpenClaw 2.0 hackathon reached its final five (Oct 1, ~4:59 PM ET), winner promised by end of day — no result in our feeds yet.
What's genuinely new: OpenClaw's ecosystem moved from skill-bazaar to reviewed marketplace in one week — vendor-distributed plugins plus an automated, model-based security review layer. That's both the strength and the question: the platform's trust layer is now another vendor's model.
Links: AIG integration · plugins (feed stories 6118 ×3)
3. The Anthropic soul fight goes religious
David Decosimo (~10:38 PM ET Oct 1, nic carter amplifying): "Anthropic secretly invited religious leaders to SF to advise them on AI safety, had them sign NDAs, but then primarily tried to convince them Claude has a soul & moral standing, all while wining & dining…" — followed today by a single-outlet item: "Anthropic vs. the pope — Claude's creators are worried the LLM could be conscious. The Catholic Church disagrees."
What's genuinely new: an institution of record is now fighting the model-welfare position from theology's side, the day after an (X-grade, unverified) report that Anthropic ran faith-leader outreach on soul/moral-standing. Full handling and watches (wire confirmation of the outreach event, Anthropic statement, whether the Church position is doctrinal) in the ai digest. Cross-ref: this is the fight Microsoft entered on record Sep 14 with its code-of-conduct clause.
4. Decision-model meme spreads: Cloudflare ships Clef, Perplexity lists its OSS haul
The category's X layer (mostly Oct 1, after the digests went to print): "Today, we're releasing two Cloudflare-trained decision models, Clef and Clef-flash" (blog, ~9:52 PM ET) — with the commentary "never seen an idea spreading so fast." Perplexity's roundup RT listed the OSS string: pplx-decider-v1-27b (85.7% avg across 11 benchmarks, "ahead of Jev"), pplx-embed-v2-context-9b-preview (SoTA contextual embeddings on ConTEB), and Lily (local…). And Arav promised to "bring down the price even further over the coming days" — ledgered (review-by Oct 8).
What's genuinely new: the meme is now vendor-ubiquitous — Cloudflare joins as distribution-owner (edge network + decision model is a routing play); Perplexity's embed-model release shows the category is spilling into adjacent model classes. Full analysis in the ai digest.
5. Economics discourse: the productivity bet and the inequality question
- Steve Rattner (NYT opinion, ~11:59 AM ET Oct 2): "Productivity growth raises per-capita incomes, but outside the dot-com boom, it's been sluggish for 50 years. A.I. is our best chance in a generation to change that." Same author day before (~9:24 PM ET Oct 1): "The S&P 500 is near a record, but if you strip out the AI companies, it's down 5% since late August" — the concentration risk frame, now charted.
- Daron Acemoglu (via Matt Stockton, ~11:09 PM ET Oct 2): "a question that also remains unasked — whether the AI boom can continue without a massive increase in inequality. A recent paper by Stijn Van Nieuwerburgh runs the numbers on how much revenue the AI industry needs to generate to pay for the buildout." The inequality question rides the same Van Nieuwerburgh revenue-math the Bain number entered the ledger from.
What's genuinely new: both sides now have their chart-craft circulating on X — the productivity optimist (Rattner) vs the distribution question (Acemoglu) vs the ex-AI market read (Rattner's own second frame). Thread layer appends, no claims to ledger beyond Arav's.
6. Brief items
- Microsoft AI: MAI-Transcribe-2-Streaming takes Artificial Analysis' streaming transcription #1, "55% faster and 60% cheaper than ElevenLabs" (Suleyman, ~4:39 PM ET Oct 1) — vendor claim + AA self-selected; watch for an ElevenLabs response.
- Accounting-acing discourse (Andrew Curran, ~3:53 AM ET Oct 2): "eighteen months ago the best models fell short of the average accountant's ~37% score. Today, models ace those same tasks" — benchmark-discourse continuity, vendor-adjacent RT chain.
- Grok group-chat vision (Andrew Curran group-chat diary, ~8:08 PM ET Oct 1): XChat group chats can see images, context-selective responses — product color, unverified claims, X-grade.
- Claude Code "You should Know" plugin (~8:29 PM ET Oct 1): scans output for missed warnings — the mods ecosystem producing genuine utilities; Anthropic praising from its own account. Mods-demo color (Sonnet 5.5 pop-up city, Opus 5.5 embossed cards, Claude Code + camera on Old Street overnight, ~8:53 PM ET) — the mod-culture gallery.
- Karpathy: "spend more time understanding LLM outputs" (~12:37 AM ET Oct 2) — tips post featuring asking models to rewrite in ASD-STE100 controlled language; practitioner color, no claim.
- Griffin (Tavus) — Roose's cautionary RT (~12:50 AM ET Oct 2): "we built the 'trick old people into handing over their bank passwords machine' from the classic sci-fi horror novel 'don't build…'" — a voice-clone product framed as satire by Roose; the underlying Tavus launch is real; agentic-attacks adjacency, X-grade, verify capabilities before weight.
Dropped
- Space RTs (Falcon Heavy NROL-97 — the NRO's first Falcon Heavy; Isaacman congratulations; Crew-13 docking; Pichai's Suncatcher-lift thank-you) — space beat, owned there; launch record already in the Oct 1 space digest.
- The $38B-subsidies counter-fight (Larry Goldberg's "$14B saved" vs WaPo analysis) — Musk-companies politics, space beat's adjacent file, not this one.
- Wolfers on healthcare-job economy; Kenneth Roth's midterm polls; Trump-checks fiscal math — macro/political, off-AI.
- vittorio's "tech tree unlocked by one guy" hero RT; Viv's "not a phone in sight" joke; Björk's reminder RT — noise.
- Photoroom's internal try-on app (Ikea/glasses verticals) — commerce CV, off-beat.
- Suraj Rana's fork-and-never-upstream prediction; 6450's platform-quality musing — discourse without a claim.
- Isomorphic Labs' 10^60 search-space RT — Sep 22 repost, outside the window and already covered in the science-claims record Oct 1.
- Artificial Analysis's Sep 30 Argon-intelligence-index RT — already digested Oct 1; the index is AA's own, vendor-adjacent.