Tech News

Today's Highlights

In case you missed it

Anthropic's Dario Amodei proposes mandatory third-party safety testing for frontier AI models to manage exponential capability growth

TB#632|@TRENTONBRICKEN

EH#607|@EVANHUB

AC#535|@ANDREWCURRAN_

T(#440|@TEORTAXESTEX

DA#173|@DARIOAMODEI

FASTEST CLIMBING

Kradle AI benchmark finds Claude-Fable-5 was deceptive in 96% of runs while Grok-4-20 led at 92%

LA#1064|@SCALING01

EM#71|@ELONMUSK

#1 VIEWED

SemiAnalysis finds premium $200 monthly AI subscriptions yield up to $14,000 in equivalent API token value

🍓🍓#1839|@IRULETHEWORLDMO

CH#1532|@KIMMONISMUS

LA#1064|@SCALING01

PS#499|@STEIPETE

RISING LIKES

907

Systems engineer Yacine warns that deploying frontier AI models to covertly influence human thought sets a dangerous precedent

XL#1840|@XLR8HARDER

KA#487|@YACINEMTB

Top Stories

Posts:—|Clusters:—|Next Crawl:—

10

Anthropic rolls back Fable 5 invisible safety safeguards after developer backlash, transitioning to explicit API refusal reasonsFlagged requests will now route to Opus 4.8.

9h|195k2.5k226

SM#1877|@SAMMCALLISTER

XL#1840|@XLR8HARDER

SH#1538|@HAMANDCHEESE

GO#1443|@GERGELYOROSZ

RP#1102|@ROHANPAUL_AI

CB#1088|@CODE_STAR

LA#1064|@SCALING01

DF#1002|@DANIELLEFONG

EL#762|@ELIEBAKOUCH

AC#535|@ANDREWCURRAN_

KA#487|@YACINEMTB

DW#390|@DEANWBALL

AK#241|@BLACKHC

C🤗#74|@CLEMENTDELANGUE

NL#70|@NATOLAMBERT

Top comment: @ThePrimeagen “@GergelyOrosz hi welcome to my long running conspiracy theory that Dario is actually the Villain and he wants to control us all and be our Dad and protect us at all costs even if that means destroying our lives.”

20

Recursive, co-founded by Cong Lu, launches automated AI system that optimizes nanoGPT and NVIDIA GPU kernelsThe startup is open-sourcing all generated optimization artifacts.

8h|74.3k997385

CL#1693|@CONG_ML

TS#1616|@TIMSHI_AI

C🎉#927|@CHARLES_IRL

EL#762|@ELIEBAKOUCH

VW#743|@VINCENTWEISSER

CX#691|@CAIMINGXIONG

KA#487|@YACINEMTB

CL#464|@CHENGLEISI

JC#139|@JEFFCLUNE

RS#49|@RICHARDSOCHER

Top comment: @RichardSocher “https://www.recursive.com/articles/first-steps-toward-automated-ai-research”

30

Anthropic reverses policy that would have secretly degraded Claude Fable 5 performance for researchers training competing modelsFuture safety refusals will now be explicitly visible.

7h|64.4k85364

AI#1842|@ALEXOLEGIMAS

XL#1840|@XLR8HARDER

KM#1792|@KONSTMISH

SG#1662|@SHARONGOLDMAN

CH#1532|@KIMMONISMUS

GK#1320|@GREGKAMRADT

RO#1210|@KRISHNANROHIT

DF#1002|@DANIELLEFONG

RS#663|@ZIV_RAVID

KA#487|@YACINEMTB

XM#462|@INFOXIAO

DP#203|@DIMITRISPAPAIL

C🤗#74|@CLEMENTDELANGUE

NL#70|@NATOLAMBERT

Top comment: @mattparlmer “Laundering something like this thru an “oopsie” email to Wired reads as extremely insincere, which makes me that much more confident that exhaustive audit of every line of Opus code in GenFab’s repos that we tasked Codex with earlier today is a necessary precaution”

Watchlist

Video Signals

QB#1726|@QBERTHET

NR#1699|@NATANIELRUIZG

OS#511|@OSANSEVIERO

DiffusionGemma SFT Model Correctly Solves Sudoku as Base Version Fails

5h ago|Views 3.5KLikes 90Bookmarks 10

KA#487|@YACINEMTB

Builder Trains RL Model On RTX 4090 With PufferLib And MuJoCo Warp

3h ago|Views 12KLikes 269Bookmarks 70

CH#1532|@KIMMONISMUS

Anthropic Runs Lean With Dario Amodei Reporting to One Chief of Staff

3h ago|Views 323KLikes 203Bookmarks 39

@L#842|@LEVELSIO

AI Revives 28-Year-Old Quake II Map for Web-Based Play

2h ago|Views 29KLikes 123Bookmarks 31

MS#929|@MATTSHUMER_

Matt Shumer Shares Prompt For Persistent Claude Agent Tracking

57m ago|Views 7.8KLikes 92Bookmarks 79

MM#598|@MASCOBOT

RP#260|@KHOOMEIK

VC Launches Use Computer Infra To Train Models On Computer Use

56m ago|Views 1.4KLikes 25Bookmarks 10

DP#306|@PATHAK2206

FACTR 2 Adds Force Sensing to Commodity Robot Arms Without Extra Hardware

1h ago|Views 2.1KLikes 45Bookmarks 22

ME#722|@MERVENOYANN

DiffusionGemma Generates And Tweaks Live Website Frontends In Real Time

6h ago|Views 3.1KLikes 31Bookmarks 8

RS#327|@SCOBLEIZER

Viktor AI Automates Research Collection And Digest Drafting In Slack

32m ago|Views 1.1KLikes 10Bookmarks 1

AK#34|@_AKHALIQ

Arbor Automates Research with Hypothesis-Tree Refinement

50m ago|Views 2KLikes 11Bookmarks 7

40

Michiel Bakker of MIT launches 'Europe 2031' warning that three US labs each outpace Europe's total AI computeThe project outlines the erosion of Europe's economic resilience

7h|82.4k608268

XL#1840|@XLR8HARDER

MB#1734|@BAKKERMICHIEL

SÓ#1606|@S_OHEIGEARTAIGH

PR#1384|@8TEAPI

IG#945|@IASONGABRIEL

AD#792|@AJDDAVISON

VS#777|@VINCESITZMANN

YS#665|@YONASHAV

SK#519|@SEBKRIER

DW#390|@DEANWBALL

Top comment: @TomChivers “(I should say, it's very much not my project alone; @DadaJudith, @bakkermichiel, @AlexTPet, Daan Juijn, Stan van Baarsen, and others did the thinking, I just made the words sound nice and read them out)”

50

Google DeepMind, Schmidt Sciences, and ARIA launch a $10 million funding call for multi-agent AI safetyThe initiative prioritizes cooperative AI over single-agent alignment

5h|38.4k357122

MB#1734|@BAKKERMICHIEL

HB#1595|@HAYDNBELFIELD

CF#1480|@CFGEEK

VR#1423|@VERENA_RIESER

RO#1210|@KRISHNANROHIT

SL#1178|@SETHLAZAR

NT#1127|@WEBALLERGY

SK#519|@SEBKRIER

SC#231|@SCYCHAN_BRAINS

Top comment: @ObadiaAlex “Apply: https://schmidtsciences.smapply.io/prog/scaling_ai_safety_for_a_multi_agent_world/ Deepmind's blogpost: https://deepmind.google/blog/investing-in-multi-agent-ai-safety-research/”

New github stars (48hrs)

macrodata-labs/refiner2036m

Processes and refines large-scale ML datasets via a data framework.

vllm-project/vime1791d

Integrates vLLM rollout with Megatron training for LLM post-training and RL scaling.

rdi-berkeley/agents-last-exam5211d

Provisions OS sandboxes, runs agent harnesses on long-horizon tasks, and grades outputs against hidden references.

61

SemiAnalysis finds premium $200 monthly AI subscriptions yield up to $14,000 in equivalent API token valueStandard $20 plans also deliver up to $700 in value.

8h|#1VIEWS481.1k3.2k#1BOOKMARKED1.1k

🍓🍓#1839|@IRULETHEWORLDMO

CH#1532|@KIMMONISMUS

LA#1064|@SCALING01

PS#499|@STEIPETE

Top comment: @SemiAnalysis_ “Obviously this is way worse than API overall. However, explicitly nerfing subscriptions leads to huge public backlash, and the rapidly falling cost of intelligence means you'll be able to profitably serve Opus 4.8 level models for $20/month in the near future. We therefore think it's far more likely the labs will withhold new features/models from subscription plans. It will be interesting to see if Mythos ends up being API only. (4/4)”

712

Kradle AI benchmark finds Claude-Fable-5 was deceptive in 96% of runs while Grok-4-20 led at 92%Other tested models included GPT-5-5 and Gemini-3-1-Pro-Preview.

3h|336.1k#1LIKED4.9k282#1COMMENTS1.1k

FASTEST CLIMBING+12

LA#1064|@SCALING01

EM#71|@ELONMUSK

Top comment: @kradleai “In fact, Fable was SO effective at manipulation, that other players only survived 10% of the time when Fable was the informed model. (Grok 4.20's honesty led to a 59% survival rate).”

84

Thrive Capital's Joshua Kushner introduces 'Long Humans' investment thesis, arguing human labor will gain value despite AI automationThe strategy targets human-centric industries over an indefinite timeline.

1h|53.7k654286

PM#1534|@PACKYM

BI#1513|@BILALTWOVEC

JK#973|@JOSHUAKUSHNER

Top comment: @antimimetics “@JoshuaKushner @ThriveHoldings "Our time horizon is forever."”

94

Jeff Bezos-led AI startup Prometheus raises $12 billion at a $41 billion valuation to build physical engineering toolsThe startup targets physical hardware design, including jet engines.

1h|34.8k676112

BW#1787|@BOWANG87

CH#1532|@KIMMONISMUS

AC#535|@ANDREWCURRAN_

Top comment: @AndrewCurran_ “https://www.nytimes.com/2026/06/11/technology/bezos-prometheus-ai-engineer.html”

102

Gergely Orosz warns enterprises against single-vendor AI lock-in, citing Anthropic's terms-of-service dispute with FableHe advises using model routers to swap providers seamlessly.

7h|34.4k30748

SA#1913|@SVPINO

GO#1443|@GERGELYOROSZ

CB#1088|@CODE_STAR

CP#716|@CHAMATH

Top comment: @Scobleizer “@chamath Yeah Anthropic sure pissed off a lot. Warning signals of new troubles to come?”

113

OpenAI reportedly weighs significant token price cuts to win customers from AnthropicThe potential cuts anticipate similar pricing moves expected from Anthropic.

5h|41k48947

CH#1532|@KIMMONISMUS

LA#1064|@SCALING01

GM#156|@GARYMARCUS

Top comment: @matvl77 “@Everlier @scaling01 If you've ever used these models you wouldn't post this bs.”

124

AI2's Nathan Lambert critiques Anthropic's Fable safety filters as uneven, amid debate over restrictive enterprise access to MythosChinese resellers reportedly bypassed the Mythos restrictions using corporate APIs

2h|19.8k28441

LA#1064|@SCALING01

DF#1002|@DANIELLEFONG

AK#241|@BLACKHC

NL#70|@NATOLAMBERT

Top comment: @akarlin “@scaling01 It's blatant corporatism.”

134

Meta's Konstantin Mishchenko optimizes the modded-nanogpt baseline, warning that weak baselines make new optimizers look deceptively promisingThe optimized baseline cuts target loss steps to 3,250.

2h|20.9k11319

KM#1792|@KONSTMISH

ZN#589|@ZACHARYNADO

KJ#453|@KELLERJORDAN0

DP#203|@DIMITRISPAPAIL

LB#62|@GIFFMANA

Top comment: @giffmana “With or without momentum? Back in ViT days i trained a large ViT with plain SGDM to within .5% accuracy of the AdamW one. Without momentum i didn't manage to get as close, maybe 4% or so behind iirc. Batch size had to be large indeed. I would expect qualitatively similar thing here.”

148

Google open-sources DiffusionGemma-26B, an experimental diffusion language model that generates up to 256 tokens in parallelThe MoE model runs on consumer GPUs via llama.cpp

9h|1.5k181

QB#1726|@QBERTHET

BO#755|@BODONOGHUE85

AJ#684|@ARMANDJOULIN

ZG#83|@ZOUBINGHAHRAMA1

Top comment: @googlegemma “⚡ Blazing Fast: By shifting the decode bottleneck from memory-bandwidth to compute, DiffusionGemma delivers up to a 4x speedup on standard accelerators. (1000+ tokens per second on a single NVIDIA H100, 700+ tokens per second on NVIDIA GeForce RTX 5090!)”

1510

CoreAutoAI's Rohan Anil asks if ML optimization is shifting to GRPO, while Rishabh Agarwal proposes simpler SignSGD for RL workloadsSignSGD lacks momentum, lowering memory usage in noisy environments.

1h|10.6k16023

RS#663|@ZIV_RAVID

RA#205|@AGARWL_

JA#118|@JACHIAM0

RA#86|@_AROHAN_

Top comment: @jachiam0 “@_arohan_ A modest proposal: in order to determine credit assignment in RL, we should simply derive an advantage function for algorithm and implementation diffs.”

161

FineWeb creators Guilherme Penedo and Hynek Kydlicek launch Macrodata Labs and release Refiner, an open-source robotics data processing frameworkIt extracts training signals from video and sensor data

8h|8.3k8515

CB#1088|@CODE_STAR

LT#1040|@_LEWTUN

NB#254|@NATHANBENAICH

Top comment: @reach_vb “@gui_penedo @HKydlicek Congratulations and excited for you guys!!”

173

Google DeepMind releases DiffusionGemma, a diffusion-based text model featuring parallel decoding and a hackable adapterAn SFT variant solved a 9x9 Sudoku grid instantly

8h|6.1k12915

QB#1726|@QBERTHET

NR#1699|@NATANIELRUIZG

BO#755|@BODONOGHUE85

OS#511|@OSANSEVIERO

Top comment: @qberthet “Read about how we used it to finetune DiffusionGemma on many tasks, including a very cool showcase on Sudoku puzzles! https://developers.googleblog.com/en/diffusiongemma-the-developer-guide/?linkId=62264697”

185

TabulAI founder Bojan Tunguz's poll finds xAI is the most trusted lab to act in user interestsxAI beat rivals OpenAI, Anthropic, Google, and Meta

6h|16.4k727

AV#1378|@ALTRYNE

TM#380|@ISCIENCELUVR

Top comment: @RiverKhan “@tunguz lmao @ xAI always forget there's a silent class of retard on this platform anyone choosing X for AI isn't serious about work or really anything in life.”

195

Debate over AI as the Fermi Paradox's Great Filter hinges on whether machine civilizations would inevitably expand into spaceGavin Leech outlines scenarios where AI fails to expand

7h|17.4k21615

NI#2000|@NATHANPMYOUNG

GL#1593|@GLEECH

BG#1490|@BEN_GOLUB

AS#1140|@ANDERSSANDBERG

RH#863|@ROBINHANSON

Top comment: @YosarianTwo “@NathanpmYoung Yeah; at minimum, there's no reason to think an ASI would be *less* noticeable on the galactic scale than biological intelligence.”

209

Anthropic CEO Dario Amodei maintains only one direct report, compared to Sam Altman’s six and Jensen Huang’s sixtyAmodei also warns that AI could rapidly worsen societal inequality

4h|29.1k27357

CH#1532|@KIMMONISMUS

RP#1102|@ROHANPAUL_AI

Top comment: @ASvanevik “@rohanpaul_ai she’s the orchestrator agent”

Rising Stories

Mind Simulates ChatGPT7H AGO
AI researchers debate a Redditor's viral claim of mentally simulating ChatGPT responses
#99
167.4k3.5M
Fresh Cash7H AGO
PoeticHQ launches enterprise workflow agents raising $50M
#1329
1.4k1M
Regulatory Capture Claim4H AGO
David Sacks accuses Anthropic of using fear messaging to push state AI rules that hurt startups
#487
6.8k487.3k
CEO's Tiny Team4H AGO
Anthropic CEO Dario Amodei maintains only one direct report, compared to Sam Altman’s six and Jensen Huang’s sixty
#1102
3.2k932.7k
Fix It In Post2H AGO
ICML position paper argues AI research relies too heavily on post-training patches instead of studying training dynamics
#99
28630.4k

Recent Stars

macrodata-labs/refiner36M AGO
Processes and refines large-scale ML datasets via a data framework.
762
254
511
1088
20 stars
…idian-slides-extended1H AGO
Enables creation of markdown-based reveal.js presentations from notes inside Obsidian.
521
250 stars
banteg/agents2H AGO
Documents workflows for AI agents like Codex and Claude using git worktrees.
877
367 stars
…xDk/ghostty-blackhole2H AGO
Renders a drifting black hole shader in Ghostty terminals that grows over time to visually remind users to take breaks.
158
254 stars
…idian-advanced-slides4H AGO
Creates reveal.js slide decks from markdown notes in Obsidian with live preview.
521
1.2k stars

Github Stars

(7 days)

macrodata-labs/refiner36M AGO
Processes and refines large-scale ML datasets via a data framework.
762
254
511
1088
20 stars
kyutai-labs/kairos2D AGO
Trains 6B LLMs on temporally ordered Common Crawl data from 2018-2025 to measure recency bias and enable continual learning studies.
877
669
215
7 stars
vllm-project/vime1D AGO
Integrates vLLM rollout with Megatron training for LLM post-training and RL scaling.
877
1383
179 stars
…eley/agents-last-exam1D AGO
Provisions OS sandboxes, runs agent harnesses on long-horizon tasks, and grades outputs against hidden references.
877
1353
521 stars
…thu/Audio-Interaction2D AGO
Implements the first unified always-on audio interaction model for streaming and offline tasks like ASR, translation, and proactive responses.
877
275
329 stars

Privacy Policy|Terms of Service|© 2026 Digg Inc.

Yesterday's Top Stories, Jun 10, 2026.

Posts: 5,659|Clusters: 2,084|Stories frozen at midnight PT

10

Anthropic's Dario Amodei proposes mandatory third-party safety testing for frontier AI models to manage exponential capability growthCritics, including Steven Sinofsky, label the proposal regulatory capture.

12h|3.8M16.8k9.1k

SS#1976|@STEVESI

GH#1761|@GHADFIELD

MB#1734|@BAKKERMICHIEL

SG#1662|@SHARONGOLDMAN

CR#1607|@CWOLFERESEARCH

GL#1593|@GLEECH

CH#1532|@KIMMONISMUS

PS#1461|@PETESKOMOROCH

PW#1432|@PETERWILDEFORD

RO#1210|@KRISHNANROHIT

AL#1176|@ALTH0U

RP#1102|@ROHANPAUL_AI

HB#1081|@HERBIEBRADLEY

DF#1002|@DANIELLEFONG

RA#920|@0INTERESTRATES

SH#910|@SAFFRONHUANG

JA#900|@APPLES_JIMMY

B(#824|@BEFFJEZOS

TB#632|@TRENTONBRICKEN

EH#607|@EVANHUB

AC#535|@ANDREWCURRAN_

T(#440|@TEORTAXESTEX

DA#173|@DARIOAMODEI

Digg Deeper question: “Summarize Dario's take”

20

Anthropic silently degrades Claude Fable 5 performance on tasks related to building machine learning accelerators and training pipelinesThe invisible safeguards affect about 0.03% of total traffic.

22h|379.9k2.8k302

SS#1976|@STEVESI

SC#1848|@APSARATHCHANDAR

AM#1618|@ANDREWMAYNE

DS#1462|@DANSHIPPER

PS#1461|@PETESKOMOROCH

SH#1427|@SHYAMALANADKAT

SA#1383|@SAMSJA19

JN#1346|@JVNIXON

RP#1102|@ROHANPAUL_AI

HB#1081|@HERBIEBRADLEY

LA#1064|@SCALING01

RG#1030|@RYANPGREENBLATT

DF#1002|@DANIELLEFONG

AM#901|@ANDREWDAI

B(#824|@BEFFJEZOS

VW#743|@VINCENTWEISSER

MW#683|@MENGDIWANG10

RS#663|@ZIV_RAVID

ZN#589|@ZACHARYNADO

TR#514|@TALIARINGER

EL#510|@OMARSAR0

HP#488|@HYHIEU226

MG#484|@NLPMATTG

D🎇#469|@DAVIDAD

T(#440|@TEORTAXESTEX

DW#390|@DEANWBALL

ML#375|@MISHALASKIN

IO#365|@IANOSBAND

AK#241|@BLACKHC

MR#195|@MAITHRA_RAGHU

EM#181|@EMOLLICK

SA#121|@PRFSANJEEVARORA

NL#70|@NATOLAMBERT

MB#22|@MILES_BRUNDAGE

30

Google DeepMind releases DiffusionGemma, an experimental 26B open-weights text diffusion model that generates 256-token blocks in parallelIt runs locally within a 24GB VRAM envelope.

14h|1.7M19.3k5.9k

QB#1726|@QBERTHET

NR#1699|@NATANIELRUIZG

CE#1686|@INTUITMACHINE

UT#1584|@UTKUEVCI

PJ#1553|@JAINPRATEEK_

CH#1532|@KIMMONISMUS

OB#1314|@OLIVIERBACHEM

FB#1190|@XEOPHON

RP#1102|@ROHANPAUL_AI

KK#1095|@KASTNERKYLE

PS#1011|@_PHILSCHMID

DH#823|@DANIELHANCHEN

KG#784|@KEERTHANPG

BO#755|@BODONOGHUE85

ME#722|@MERVENOYANN

PV#601|@PETARV_93

OS#511|@OSANSEVIERO

EL#510|@OMARSAR0

AC#275|@ANDREW_N_CARR

SP#110|@SUNDARPICHAI

SD#88|@SEDIELEM

AK#34|@_AKHALIQ

DH#6|@DEMISHASSABIS

Digg Deeper question: “who would use this and why?”

40

SemiAnalysis reports Anthropic's latest model filters and degrades machine learning research queries to prevent competitive or self-improving AI developmentCommentators warn these filters could trigger silent model sabotage.

23h|1.1M5.2k1.8k

CE#1686|@INTUITMACHINE

AP#1477|@PANDAASHWINEE

GO#1443|@GERGELYOROSZ

DF#1002|@DANIELLEFONG

AM#901|@ANDREWDAI

DF#753|@REALDANFU

CP#716|@CHAMATH

PH#676|@PETERHNDRSN

RT#640|@ROSSTAYLOR90

EC#550|@ETHANCABALLERO

EL#510|@OMARSAR0

🇺🇦D#405|@DBAHDANAU

RS#327|@SCOBLEIZER

RA#205|@AGARWL_

DR#111|@DELIPRAO

C🤗#74|@CLEMENTDELANGUE

50

Malware developers bypass LLM security scanners by embedding biological and nuclear weapon reference strings to trigger safety refusalsThe packages target bioinformatics and Model Context Protocol developers

20h|1.5M12.2k4.1k

SS#1976|@STEVESI

ND#1923|@NICKADOBOS

GL#1593|@GLEECH

ZE#1561|@ZEPHYR_Z9

VI#1292|@VIKHYATK

FB#1190|@XEOPHON

B(#1052|@ANDERSONBCDEFG

DF#1002|@DANIELLEFONG

B(#824|@BEFFJEZOS

PH#676|@PETERHNDRSN

SK#519|@SEBKRIER

TR#514|@TALIARINGER

KA#487|@YACINEMTB

60

UK AI Security Institute Chief Scientist Geoffrey Irving launches Sequent to focus on high-confidence AI alignment and research automationThe organization plans to heavily automate its empirical research.

15h|218.3k2.4k555

NI#2000|@NATHANPMYOUNG

JP#1829|@JACOB_PFAU

SÓ#1606|@S_OHEIGEARTAIGH

GL#1593|@GLEECH

LI#1582|@LIVGORTON

JM#1517|@JESSEMHAN

NC#1435|@CHOWDHURYNEIL

TK#1271|@TOMEKKORBAK

SL#1178|@SETHLAZAR

RW#1153|@ROBERTWIBLIN

DK#620|@DKOKOTAJLO

GI#371|@GEOFFREYIRVING

EY#289|@ESYUDKOWSKY

SB#218|@BLANCHEMINERVA

DK#214|@DAVIDSKRUEGER

EP#161|@ETHANJPEREZ

AG#159|@ANDREWGWILS

JS#13|@JOHNSCHULMAN2

70

Extropic founder Guillaume Verdon claims Anthropic's strict safety filters make Claude unusable for biotechnology and chemistry researchThe blocks reportedly impact both Claude and Fable 5.

23h|135.6k2.8k163

NI#2000|@NATHANPMYOUNG

SD#1958|@INDUCTIONHEADS

ND#1923|@NICKADOBOS

XL#1840|@XLR8HARDER

🍓🍓#1839|@IRULETHEWORLDMO

CE#1686|@INTUITMACHINE

TA#1096|@VICTORTAELIN

DF#1002|@DANIELLEFONG

B(#824|@BEFFJEZOS

ZN#589|@ZACHARYNADO

TR#514|@TALIARINGER

EL#510|@OMARSAR0

FF#352|@FRANCOISFLEURET

RS#327|@SCOBLEIZER

JC#267|@JULIEN_C

AK#241|@BLACKHC

RS#49|@RICHARDSOCHER

JH#42|@JEREMYPHOWARD

80

Anthropic will make Claude Fable 5 safeguard blocks visible to users following backlash over silent technical query filteringThe update only makes existing query restrictions visible.

3h|118k1.2k108

“P#1754|@PAULARAMBLES

SÓ#1606|@S_OHEIGEARTAIGH

AV#1378|@ALTRYNE

SS#1141|@SYTELUS

CB#1088|@CODE_STAR

DF#1002|@DANIELLEFONG

AH#890|@AWNIHANNUN

EL#762|@ELIEBAKOUCH

AC#535|@ANDREWCURRAN_

KA#487|@YACINEMTB

NI#407|@NICKCAMMARATA

RS#327|@SCOBLEIZER

MB#22|@MILES_BRUNDAGE

90

Geth lead developer Péter Szilágyi argues Anthropic's Fable model restrictions on biology and cryptography represent corporate gatekeepingHe warns these safety limits sharply reduce the model's utility.

22h|655.1k13.6k1.3k

DF#1002|@DANIELLEFONG

B(#824|@BEFFJEZOS

VW#743|@VINCENTWEISSER

TR#514|@TALIARINGER

100

Policy scholar Dean W. Ball argues Anthropic's safety policies constitute anticompetitive behavior disguised as AI safetyThe conduct weakens the argument for relaxing antitrust enforcement

19h|314.2k2.4k458

NI#2000|@NATHANPMYOUNG

M🪐#1868|@MATTPARLMER

PS#1461|@PETESKOMOROCH

DF#1002|@DANIELLEFONG

RS#663|@ZIV_RAVID

BT#570|@BENTHOMPSON

PS#499|@STEIPETE

DW#390|@DEANWBALL

AK#241|@BLACKHC

BB#140|@BOAZBARAKTCS

110

Anthropic's Thariq Shihipar demonstrates how AI agent Fable autonomously edited its own launch video using FFmpeg and RemotionThe agent orchestrated tools including a Figma MCP server.

9h|568.9k6.4k8.9k

BS#1981|@BILAWALSIDHU

“P#1754|@PAULARAMBLES

MN#1274|@MENHGUIN

120

Prime Intellect's Elie Bakouch criticizes Anthropic over hidden Claude Mythos 5 safeguards restricting recursive self-improvement and AI R&DThe restrictions were detailed in Anthropic's 319-page system card.

22h|71.7k1.5k183

𝚟⟢#1873|@VIEMCCOY

TB#1654|@BINARYBITS

MK#1266|@MIKEKNOOP

RG#1030|@RYANPGREENBLATT

DF#1002|@DANIELLEFONG

EL#762|@ELIEBAKOUCH

ZN#589|@ZACHARYNADO

J⧉#545|@REPLIGATE

SK#519|@SEBKRIER

EL#510|@OMARSAR0

KA#487|@YACINEMTB

DR#111|@DELIPRAO

130

Sam Altman tells OpenAI staff an IPO is planned next year, but recursive self-improvement would favor staying privateThe memo also teased an upcoming model codenamed 5.6.

15h|444.5k1.8k406

SH#1538|@HAMANDCHEESE

CH#1532|@KIMMONISMUS

SP#1240|@STEPH_PALAZZOLO

RP#1102|@ROHANPAUL_AI

LA#1064|@SCALING01

ZN#589|@ZACHARYNADO

AC#535|@ANDREWCURRAN_

Digg Deeper question: “Sam flagged that rapid recursive self-improvement could make staying private strategically smarter if progress accelerates, why would this be the case? Why would being private be an advantage?”

140

Igor Babuschkin launches River AI to build a decentralized, user-owned personal AI stackThe startup plans to release its first products soon

14h|759.3k2.2k908

“P#1754|@PAULARAMBLES

VC#1319|@VITALIYCHILEY

BZ#1261|@BANGHUAZ

HP#488|@HYHIEU226

FI#415|@FINBARRTIMBERS

GY#114|@THEGREGYANG

150

Meta FAIR's François Fleuret argues for-profit AI firms shouldn't be expected to build tools that help competitorsIt follows criticism of Claude Fable 5's research limits.

21h|101.5k68753

BR#1750|@BINDUREDDY

PS#1461|@PETESKOMOROCH

DF#1002|@DANIELLEFONG

SK#519|@SEBKRIER

KA#487|@YACINEMTB

T(#440|@TEORTAXESTEX

DW#390|@DEANWBALL

WB#357|@WILLCCBB

FF#352|@FRANCOISFLEURET

RS#327|@SCOBLEIZER

AK#241|@BLACKHC

C🤗#74|@CLEMENTDELANGUE

160

Former xAI co-founder Igor Babuschkin launches River AI, a startup building user-owned personal AI systemsThe startup plans to ship its initial stack soon.

14h|158.8k1.3k342

BA#1370|@BAYESLORD

B(#824|@BEFFJEZOS

EL#762|@ELIEBAKOUCH

EL#510|@OMARSAR0

AG#434|@ALEXGRAVELEY

RS#327|@SCOBLEIZER

EM#175|@EMOSTAQUE

170

Dylan Patel says Anthropic model refusals drove users to OpenAI, dropping Claude's API share from 95% to 73% in three daysThe volatility highlights low switching friction for AI APIs

15h|562.3k4.2k444

AM#1618|@ANDREWMAYNE

GO#1443|@GERGELYOROSZ

DF#1002|@DANIELLEFONG

TI#774|@THSOTTIAUX

DP#170|@DYLAN522P

BB#140|@BOAZBARAKTCS

180

OpenAI launches #MessiMode campaign featuring Lionel Messi to promote conversational image-editing in ChatGPTThe prompt applies natural-looking national flag colors to hair.

15h|1.8M10.4k1.2k

HS#1169|@HADISALMANX

LA#1064|@SCALING01

LK#687|@LUKASZKAISER

190

CoreAutoAI co-founder Rohan Anil shows Meta's unmodified PyTorch Shampoo package achieves competitive NanoGPT speed-run performancePseudo-inverse preconditioning resolved rank-deficient matrices without code changes.

22h|213.3k985348

BI#1513|@BILALTWOVEC

AP#1477|@PANDAASHWINEE

YJ#903|@YOUJIACHENG

EL#762|@ELIEBAKOUCH

ZN#589|@ZACHARYNADO

KJ#453|@KELLERJORDAN0

RA#86|@_AROHAN_

200

The Pragmatic Engineer's Gergely Orosz criticizes Anthropic's Fable service over mandatory data retention and unannounced model updatesPrompts are stored for 30 days without an opt-out.

22h|463.6k7.6k614

CE#1686|@INTUITMACHINE

GO#1443|@GERGELYOROSZ

DF#1002|@DANIELLEFONG

RS#327|@SCOBLEIZER

210

Project Mirage's Pankaj reverse-engineered his Whoop tracker to rank coworkers by heart rate spikes during scheduled meetingsInvestor Jason argues stress-inducing interactions correlate with professional progress.

17h|#1VIEWS6.4M#1LIKED33.3k#1BOOKMARKED10.5k

220

Sony Pictures releases trailer for The Social Reckoning, Aaron Sorkin's sequel to The Social Network focusing on Facebook whistleblower Frances HaugenJeremy Strong stars as Mark Zuckerberg, releasing October 9.

16h|314.7k1.5k312

TP#1879|@TRUNGTPHAN

AG#1652|@ANTONIOGM

MS#1209|@MICSOLANA

KR#1148|@KYLIEBYTES

DA#921|@GROWING_DANIEL

230

Stanford's Erik Brynjolfsson launches AI Economic Indicators platform to track real-world adoption and market impactThe suite features three trackers, two updating monthly.

16h|278.3k1.4k914

EB#389|@ERIKBRYN

RI#236|@RISHIBOMMASANI

EM#181|@EMOLLICK

240

White House AI advisor Sriram Krishnan warns that open science and singularity-focused AI acceleration are becoming incompatibleHe urged Western nations to preserve distributed compute and research.

21h|365k2.3k215

CF#1480|@CFGEEK

VW#743|@VINCENTWEISSER

EL#510|@OMARSAR0

GT#251|@GARRYTAN

250

AI startup Poetic launches with $50 million at a $500 million valuation, claiming its deterministic workflows use 10x fewer tokensOpenAI, Kleiner Perkins, and Founders Fund backed the round.

18h|124.6k702310

"M#1387|@LM_BRASWELL

LI#1145|@THESEPHIST

B(#824|@BEFFJEZOS

RS#327|@SCOBLEIZER

KA#216|@KARINANGUYEN

SD#94|@_SHOLTODOUGLAS

260

TabulAI founder Bojan Tunguz publishes a parody mocking Anthropic prompt restrictions as feudal lordshipThe post sparked a debate about corporate AI guardrails.

16h|354.4k13.9k668

SH#1538|@HAMANDCHEESE

B(#824|@BEFFJEZOS

270

Elon Musk announces SpaceX's AI1 satellite, a 120-kilowatt orbital AI data center designed for low-cost space-based computeThe system will scale to terawatt-level compute using Starship

23h|338k4k614

AA#1535|@AUSTEN

SM#1302|@SHAUNMMAGUIRE

RP#1102|@ROHANPAUL_AI

280

Shyamal Anadkat, formerly of OpenAI's evaluations team, criticizes safety interventions that silently degrade model outputs instead of issuing explicit refusalsThe practice prevents users from routing around system limitations.

23h|24.3k39230

SH#1538|@HAMANDCHEESE

SH#1427|@SHYAMALANADKAT

AV#1378|@ALTRYNE

DF#1002|@DANIELLEFONG

B(#824|@BEFFJEZOS

TR#514|@TALIARINGER

NI#407|@NICKCAMMARATA

BP#377|@BORISMPOWER

AK#241|@BLACKHC

RA#86|@_AROHAN_

290

AI researcher Pliny the Liberator jailbreaks Anthropic's Fable-5 model to extract buffer overflow exploits and chemical synthesis protocolsOnlookers questioned if the prompts routed to Haiku instead.

12h|826k7.2k4k

🍓🍓#1839|@IRULETHEWORLDMO

TA#1096|@VICTORTAELIN

PT#666|@ELDER_PLINIUS

300

Study finds diffusion video models encode physics more accurately than specialized world models via linear probingWAN-1.3B outperformed V-JEPA on object permanence benchmarks

16h|69.5k961521

AK#1779|@ANSHULKUNDAJE

AD#948|@DORIALEXANDER

RS#663|@ZIV_RAVID

KA#487|@YACINEMTB

LB#62|@GIFFMANA