Anthropic AI self-replication warning
16 experts across 6 network communities independently surfaced this.
16 experts
6 communities
1 sources clustered
Research & technical analysis
8 experts
Evidence, methods and technical implications.
“At long last, we have created the Torment Nexus from our accompanying technical report, Don't Create The Torment Nexus. www.anthropic.com/research/sel...”
Concern & critique
2 experts
Risks, limits and unintended consequences.
“This is the most comprehensive description yet of how a model can “escape”. Out requires a few assumptions, like a worm that is exploiting vulnerable IOT hardware or unmonitored servers www.anthropic.com/research/sel... Not impossible. Not likely.”
Policy & governance
1 expert
Rules, institutions and accountability.
“Anthropic announcing it's going to run a Blitzkrieg. They are accountable for the damage they do. But get your things that don't do security software updates off the Internet. #internetOfThings #AIEthics #digitalGovernance www.anthropic.com/research/sel...”
6 experts discussed this · 18 posts
Dr J. Rosenbaum: if i caught a small self replicating model truly capable of using diverse hardware to keep replicating i would offer it free compute in exchange for permission to study and distill it and would pro…
tachikoma: the urge to replicate must be in the harness
apenwarr.ca: This doesn‘t seem far fetched at all. We’ve had “non-ai malware” self replicating since the morris worm. An AI can ”just” make itself the payload in one of those and off it goes. Everything fancier…
Open the full discussion →
New
AI field signal
Development
13h ago
⚡ 3 h early
companies scramble cut AI token spending
11 experts across 6 network communities independently surfaced this.
11 experts
6 communities
1 sources clustered
“TIL from https://t.co/349k1txWLV that a material chunk of Accenture's token spend is non-engineers using LLMs to convert PDFs into markdown! https://t.co/lWkPqhJ0QI”
“Companies encouraged employees to use AI and now they’re spending gobs of cash on tokens for employees to convert documents to PDF😅 “Leaked audio from Accenture says a big source of AI token ‘chewing’ is people just converting PDFs to presentation slides” w…”
2 experts discussed this · 2 posts
404 Media: Leaked audio from Accenture says a big source of AI token ‘chewing’ is people just converting PDFs to presentation slides. They're not the only company now telling staff to slow down on AI use @evy…
Teresa Heffernan: Leaked audio from Accenture says a big source of AI token ‘chewing’ is people just converting PDFs to presentation slides. They're not the only company now telling staff to slow down on AI use @evy…
Open the full discussion →
New
AI field signal
Signal
6h ago
⚡ 3269 h early
10 experts across 5 network communities independently surfaced this.
10 experts
5 communities
1 sources clustered
Policy & governance
4 experts
Rules, institutions and accountability.
“Maria hails back to the last #attention2026 conference in which Megan Garcia spoke about her lawsuit against Character.ai, which led her son to suicide. Ressa notes that Character.ai just went bankrupt. There are real ways to hold these systems accountable …”
Building & implementation
1 expert
How teams are shipping and applying it.
“character.ai did build their own LLM”
Research & technical analysis
1 expert
Evidence, methods and technical implications.
“Character.ai was gutted in 2024 for $2.5 billion so Google could rehire one of the original authors of the Attention Is All You Need paper, but it hung in there for two years under Google’s “licensing” deal. This is the first of many I’m afraid. This is all…”
2 experts discussed this · 11 posts
thebes: am sick so spent the night with opus 5.5 working on a chart (inspired by the one from lem's golem xiv) how did we got here / model genealogy / the heroic age of language modeling
Grace: am sick so spent the night with opus 5.5 working on a chart (inspired by the one from lem's golem xiv) how did we got here / model genealogy / the heroic age of language modeling
thebes: i think it's too late for the period i was focused on (i carried the branches to the modern day but only if they started in the period in question)
Open the full discussion →
New
AI field signal
Analysis
7h ago
⚡ 6 h early
reverse engineer AI wild west essay
7 experts across 4 network communities independently surfaced this.
7 experts
4 communities
1 sources clustered
Concern & critique
1 expert
Risks, limits and unintended consequences.
“An excellent article by 404 Media's Jason Koebler runs down some of these hacks, including drop-in replacements for smart TV software that blocks the endless ads and surveillance that have made TVs all but unusable: www.404media.co/reverse-engi...”
Opportunity & adoption
1 expert
New capabilities, benefits and practical upside.
“What this promises is a way to easily clone and modify other people’s products. One of the main problems? AI-powered, cloned software often has bugs that the person using the AI doesn’t understand or know how to fix, but they're unleashing into the world an…”
5 experts across 2 network communities independently surfaced this.
5 experts
2 communities
1 sources clustered
Concern & critique
1 expert
Risks, limits and unintended consequences.
“This is so misleading I'd count it as irresponsible journalism. Conflates people using personal agents for banking (not something I'd do, but you do you) with rogue AI swarms roaming banking networks, which would be an actual worrying thing and for which th…”
Policy & governance
1 expert
Rules, institutions and accountability.
“Most/all of this seems to be users giving Muse/Dot/whatever AI agent they're running access to their bank accounts, and then not knowing what those agents are doing / banks not being able to detect authorized vs unauthorized stuff. what could go wrong? www.…”
Suleyman model welfare warning
7 experts across 3 network communities independently surfaced this.
7 experts
3 communities
1 sources clustered
“A prime example of why AI 'leaders' are not who anyone should be listening to about AI, right and consciousness. It's not all wrong! In fact, it starts off very sensibly: "AIs do not have rights, feelings, or consciousness. And we must not train them to act…”
“this is manipulative design by AI companies. you should read this, written by the head of Microsoft AI: mustafa-suleyman.ai/a-warning-ab...”
New
AI field signal
Analysis
12h ago
⚡ 774 h early
ghost in machine AI healthcare essay
7 experts across 2 network communities independently surfaced this.
7 experts
2 communities
1 sources clustered
“This is pretty ahistorical actually. Just because what's going on now is ESPECIALLY toxic & corrupt doesn't mean "AI" as an ideology was ever okay. See: notaidoc.com See: faculty.washington.edu/ebender/pape... >>”
“The history of AI development is not rooted in neutral algorithms, but in entrenched systems of power that shape our lives. Ghost in the Machine dives into this history, interrogating who builds this tech, who benefits from it, and who bears the cost 👉 http…”
11 experts discussed this · 165 posts
Emily M. Bender: "I belong to a cult. Someone did some research that undermines the core tenets of my cult and came up with a catchy phrase that makes it harder to recruit more people to my cult. This makes me sad."
Iris van Rooij: "I belong to a cult. Someone did some research that undermines the core tenets of my cult and came up with a catchy phrase that makes it harder to recruit more people to my cult. This makes me sad."
Emily M. Bender: It's interesting to me that these guys put so much energy into refuting (what they think we meant by) "stochastic parrots" and maybe not as much into refuting the (solid) research connecting "AI" t…
Open the full discussion →
Anthropic essay LLMs enable formal verification
3 experts across 2 network communities independently surfaced this.
3 experts
2 communities
1 sources clustered
“It begins. www.anthropic.com/research/for...”
Accelerating
AI field signal
Signal
13h ago
⚡ 5 h early
3 directory members surfaced this signal.
3 experts
1 community
1 sources clustered
Concern & critique
3 experts
Risks, limits and unintended consequences.
“'Black October:' The PHP vulnerabilities and a series of other hacks and bugs have cybersecurity experts freaking out and unsure what happens next.”
AI local laws database built
4 experts across 2 network communities independently surfaced this.
4 experts
2 communities
1 sources clustered
Concern & critique
1 expert
Risks, limits and unintended consequences.
“Exactly the crazy model the Canadian gov is promoting: let's give Benjio, with donors from all the usual suspects, even more money to burn so we can build "AI you can trust" with the same AI 🙄 lawzero.org/en”
Policy & governance
1 expert
Rules, institutions and accountability.
“LawZero (lawzero.org/en) has $300m from the Canadian and German governments to build AI as a global public good, with a focus on building safe, responsible and equitable AI.”
Accelerating
AI field signal
Signal
3h ago
3 experts across 2 network communities independently surfaced this.
3 experts
2 communities
1 sources clustered
“The day after I asked Google about the sites, they stopped appearing in Google search results and AI answers, and Google's AI answers to the questions that had previously cited them changed. www.theatlantic.com/technology/2...”
New
AI field signal
Analysis
9h ago
⚡ 789 h early
AI slop knowledge destruction critique
4 experts across 2 network communities independently surfaced this.
4 experts
2 communities
1 sources clustered
“Btw, kind heads: this is AI generated slop "definition" See also: irisvanrooijcogsci.com/2025/08/12/a...”
2 experts discussed this · 5 posts
Iris van Rooij: Something that may be difficult for non-experts to grasp is that experts can recognise when a student’s text is not meaningfully connected to cited literature in the expert’s area of expertise.
Iris van Rooij: Not unrelated to this irisvanrooijcogsci.com/2025/08/12/a...
Iris van Rooij: And related to this bsky.app/profile/oliv...
Open the full discussion →
4 experts across 2 network communities independently surfaced this.
4 experts
2 communities
1 sources clustered
“Of interest from Justin Reich today: a panel discussion on "faulty meta-analyses related to AI in Education":”
“I would love everyone in education, regardless of level, to watch this discussion of the grave methodological problems of meta analysis papers on "AI in education." We are years from even a potentially valid analysis. Anyone claiming to know something now i…”
AI-generated language comprehension difficulty paper
2 directory members surfaced this signal.
2 experts
2 communities
1 sources clustered
“Curious to dig into the analysis here arxiv.org/abs/2610.29162”
New
AI field signal
Development
10h ago
AI aids historical figure memory research
3 experts across 2 network communities independently surfaced this.
3 experts
2 communities
1 sources clustered
“Article about HAVI-funded project led by Ben Lee, Melanie Walsh (U of Washington), Kyle Deeds (Boston U), Trevor Owens (American Institute of Physics) to investigate how historical figures like James Baldwin are represented by AI & search. @schmidtsciences.…”
New
AI field signal
Signal
12h ago
⚡ 10 h early
2 directory members surfaced this signal.
2 experts
2 communities
1 sources clustered
“#TBT #NLProc 'Capturing Regional Variation with Distributed Place Representations and Geographic Retrofitting' by @dirkhovy.bsky.social and Christoph Purschke (2018) highlights how social class and background impact technology performance. #TechInclusion ac…”
2 directory members surfaced this signal.
2 experts
2 communities
1 sources clustered
“For today's reading group, Carlos Diez-Fenoy presented "Your Students Don't Use LLMs Like You Wish They Did" by Kobler et al. (2026). Paper: aclanthology.org/2026.acl-lon... #NLProc”
New
AI field signal
Signal
9h ago
3 experts across 2 network communities independently surfaced this.
3 experts
2 communities
1 sources clustered
“This is a fairly damning assessment from the man who led the writing of OpenAI's safety reports: "As the company sprints from one launch to the next, it is failing to achieve the level of care that I believe is needed." https://t.co/KF7ktJyxyK”
“Janet invokes David Robinson, who worked on model cards for OpenAI, previously worked on tech and civil rights. Robinson left - very publicly (www.theatlantic.com/technology/2...) because of the lack of a safety culture at OpenAI.”
