AI News

It’s a new day at Digg. Fresh stories are clustering now. 65% complete.

Top Stories

Posts:—|Clusters:—|Next Crawl:—

10

Anthropic reportedly restricts machine learning research queries and blocks tasks classified as self-improving for other modelsThe filters degrade performance on paid advanced programming queries.

3h|33.7k25648

CE#1596|@INTUITMACHINE

PH#630|@PETERHNDRSN

RT#598|@ROSSTAYLOR90

EC#519|@ETHANCABALLERO

C🤗#67|@CLEMENTDELANGUE

21

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.

3h|20k55029

XL#1671|@XLR8HARDER

B(#850|@BEFFJEZOS

FF#330|@FRANCOISFLEURET

RS#323|@SCOBLEIZER

31

Jeremy Howard, fast.ai co-founder, accuses Anthropic of silently sabotaging scientific experiments to protect its proprietary leadHoward mockingly referred to the AI startup as "Sophanthropic.

3h|21.7k44745

B(#850|@BEFFJEZOS

RS#323|@SCOBLEIZER

AK#228|@BLACKHC

JH#38|@JEREMYPHOWARD

Watchlist

Video Signals

CH#1448|@KIMMONISMUS

AI Prompt Generates Fable-Style 3D Forest in Three.js

1h ago|Views 68KLikes 124Bookmarks 24

B(#850|@BEFFJEZOS

Meme Mocks Closed AI Models With Apple's 1984 Ad

3h ago|Views 5.2KLikes 72Bookmarks 13

B(#850|@BEFFJEZOS

Beff Jezos Mocks Claude With Skyrim Guard Meme For ML Research

2h ago|Views 3.7KLikes 47Bookmarks 5

GA#712|@GABRIEL1

Legora Opens Offices In Madrid, Milan, Paris And London Engineering Hub

3h ago|Views 12KLikes 58Bookmarks 16

KD#932|@CSPROFKGD

Ego-MC-Bench Benchmark Tests AI Mistake Correction in Real-Time Tasks

13h ago|Views 946Likes 9Bookmarks 5

RV|@RONALD_VANLOON

Humanoid Robot Xiaoliang Cleans Homes and Greets Elderly Users

3h ago|Views 299Likes 2Bookmarks 2

RV|@RONALD_VANLOON

Heyman Tech Unveils Waterproof Humanoid Hand for Extreme Underwater Depths

22m ago|Views 133Likes 1Bookmarks 0

RV|@RONALD_VANLOON

Field AI Deploys Adaptive Humanoid Robots for Safe Operation in Unknown Environments

1h ago|Views 145Likes 1Bookmarks 1

RV|@RONALD_VANLOON

EIR Technology Unveils Humanoid Robot With Advanced Mobility Features

2h ago|Views 159Likes 1Bookmarks 0

42

Anthropic quietly restricts Claude Fable 5 from helping build ML accelerators and pretraining pipelinesThe silent limits affect roughly 0.03 percent of traffic.

2h|2.9k380

DS#1427|@DANSHIPPER

LA#975|@SCALING01

B(#850|@BEFFJEZOS

ML#352|@MISHALASKIN

IO#341|@IANOSBAND

51

Deedy Das of Menlo Ventures praises Claude Fable's design capabilities amid reports of developer backlashOutputs included a Pokémon clone and 3D graphics.

1h|29.8k401266

DE#363|@DEEDYDAS

RS#323|@SCOBLEIZER

New github stars (48hrs)

vllm-project/vime1915h

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

datacurve-ai/deep-swe71815h

Benchmarks frontier coding agents on 113 long-horizon tasks from active open-source repos with isolated environments and program verifiers.

xzf-thu/Audio-Interaction2831d

Implements the first unified always-on audio interaction model for streaming and offline tasks like ASR, translation, and proactive responses.

kyutai-labs/kairos71d

Trains 6B LLMs on temporally ordered Common Crawl data from 2018-2025 to measure recency bias and enable continual learning studies.

roboflow/supervision42.9k1d

Supplies unified Detections API, annotators, trackers, and dataset converters for computer vision pipelines.

61

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.

1h|8.2k13732

EL#706|@ELIEBAKOUCH

KJ#426|@KELLERJORDAN0

RA#79|@_AROHAN_

7NEW

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.

54m|5k420

FF#330|@FRANCOISFLEURET

RS#323|@SCOBLEIZER

AK#228|@BLACKHC

81

Extropic founder Guillaume Verdon argues a resurfaced 2023 Anthropic pitch deck shows major AI firms restricting market accessThe document sparked viral debates about AI industry gatekeeping

3h|7.5k779

B(#850|@BEFFJEZOS

RS#323|@SCOBLEIZER

91

Jeremy Howard concedes his proposal to restrict frontier AI labs from recursive self-improvement is ineffectiveHoward remains opposed to artificial slowdowns of AI development

2h|2k200

AK#228|@BLACKHC

JH#38|@JEREMYPHOWARD

101

EleutherAI's Stella Biderman argues CSAM prevention is the only clear scenario where closed-source models are safer than open-sourceShe also compiled documentation showing WormGPT was a fraud

3h|50661

FB#1117|@XEOPHON

SB#207|@BLANCHEMINERVA

111

Google DeepMind's Andreas Kirsch argues open-source AI will soon match proprietary models like Mythos, drawing pushback from Florian BrandBrand demanded concrete evidence rather than optimistic projections.

2h|781140

FB#1117|@XEOPHON

AK#228|@BLACKHC

121

Princeton's Mengdi Wang claims AI research requires massive compute, but Pasquale Minervini argues single-GPU projects can achieve state-of-the-art resultsMinervini's outstanding paper runs overnight on a single GPU

1h|41471

PM#710|@PMINERVINI

MW#638|@MENGDIWANG10

131

Andreas Kirsch warns of "social panic" over a perceived Claude nerf as Extropic founder Guillaume Verdon defends the backlashSome researchers claim the updated model intentionally introduces bugs.

3h|492200

B(#850|@BEFFJEZOS

AK#228|@BLACKHC

Recent Stars

extend-hq/ui4H AGO
UNKNOWN.
1170
203 stars
…rcks/mythos-adventure5H AGO
A PS1-style 3D adventure game rendered in your terminal — Rust + Ratatui, played through a chat box
1784
1 stars
mvanhorn/agentcookie13H AGO
Syncs browser cookies, tokens, and auth state between Macs over Tailscale for remote AI agent runtimes.
1171
339 stars
2389-research/thrifty13H AGO
Delegates Claude Code tasks by having a strong model plan sprints with contracts then cheap models execute and self-verify against gates.
1133
1 stars
ljang0/iOSWorld14H AGO
Benchmarks LLM agents on iPhone Simulator using 26 SwiftUI apps, 133 tasks, and rubrics.
933
2 stars

Github Stars

(7 days)

kyutai-labs/kairos1D AGO
Trains 6B LLMs on temporally ordered Common Crawl data from 2018-2025 to measure recency bias and enable continual learning studies.
862
631
203
7 stars
vllm-project/vime15H AGO
Integrates vLLM rollout with Megatron training for LLM post-training and RL scaling.
862
1266
19 stars
datacurve-ai/deep-swe15H AGO
Benchmarks frontier coding agents on 113 long-horizon tasks from active open-source repos with isolated environments and program verifiers.
862
1970
718 stars
…thu/Audio-Interaction1D AGO
Implements the first unified always-on audio interaction model for streaming and offline tasks like ASR, translation, and proactive responses.
862
260
283 stars
roboflow/supervision1D AGO
Supplies unified Detections API, annotators, trackers, and dataset converters for computer vision pipelines.
862
203
42.9k stars

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

Yesterday's Top Stories, Jun 9, 2026.

Posts: 6,333|Clusters: 1,715|Stories frozen at midnight PT

10

Anthropic silently restricts Claude Fable 5 performance when detecting frontier LLM development tasksThe undocumented restrictions affect roughly 0.03% of overall traffic.

13h|2.5M20.2k2.5k

AM#1930|@GRAVICLE

ND#1871|@NICKADOBOS

AP#1680|@ALEX_PEYS

XL#1671|@XLR8HARDER

TM#1646|@BANBURISMUS_

JA#1546|@JULIUSADML

ZE#1471|@ZEPHYR_Z9

CH#1448|@KIMMONISMUS

NS#1411|@NABEELQU

WH#1409|@NREHIEW_

1A#1393|@1A3ORN

PS#1385|@RECURSEPARADOX

BI#1366|@BILALTWOVEC

PR#1330|@8TEAPI

BA#1291|@BAYESLORD

LC#1248|@LCASTRICATO

AV#1245|@ALTRYNE

RO#1214|@KRISHNANROHIT

FB#1117|@XEOPHON

SS#1061|@SYTELUS

CB#1004|@CODE_STAR

ZW#851|@ZHAORAN_WANG

B(#850|@BEFFJEZOS

🇺🇦A#785|@SKIMINOK

X(#735|@XUANALOGUE

VW#696|@VINCENTWEISSER

RS#617|@ZIV_RAVID

HE#577|@HEINRICHKUTTLER

ZN#551|@ZACHARYNADO

EC#519|@ETHANCABALLERO

AC#518|@ANDREWCURRAN_

KA#488|@YACINEMTB

T(#421|@TEORTAXESTEX

XE#407|@XWANG_LK

DW#390|@DEANWBALL

AK#228|@BLACKHC

SB#207|@BLANCHEMINERVA

DP#193|@DIMITRISPAPAIL

GM#157|@GARYMARCUS

FH#145|@FHUSZAR

BB#133|@BOAZBARAKTCS

DR#102|@DELIPRAO

C🤗#67|@CLEMENTDELANGUE

NL#64|@NATOLAMBERT

DR#56|@ROYDANROY

LB#55|@GIFFMANA

MB#20|@MILES_BRUNDAGE

20

Anthropic launches Claude Fable, a Mythos-class model that scored 91 on Every's senior-engineer benchmarkThe model costs twice as much as Claude Opus.

1d|4.3M33.3k4.7k

🍓🍓#1713|@IRULETHEWORLDMO

BR#1599|@BINDUREDDY

AM#1489|@ANDREWMAYNE

ZE#1471|@ZEPHYR_Z9

CH#1448|@KIMMONISMUS

DS#1427|@DANSHIPPER

PW#1345|@PETERWILDEFORD

PR#1330|@8TEAPI

GB#1275|@GAVINSBAKER

AV#1245|@ALTRYNE

RP#1031|@ROHANPAUL_AI

LA#975|@SCALING01

AH#802|@AWNIHANNUN

EL#706|@ELIEBAKOUCH

PD#653|@PMDDOMINGOS

AC#518|@ANDREWCURRAN_

AA#440|@ALEXALBERT__

RS#323|@SCOBLEIZER

JC#253|@JULIEN_C

GM#157|@GARYMARCUS

DR#102|@DELIPRAO

30

Anthropic restricts its Mythos-based models on machine learning research and bans open-source developer @willccbbThe restrictions use steering vectors to throttle research tasks.

12h|160.3k3k271

NI#1900|@NATHANPMYOUNG

SH#1490|@HAMANDCHEESE

MG#1350|@MEGAMOR2

TH#1286|@TIMHWANG

SA#1266|@SAMSJA19

FB#1117|@XEOPHON

SS#1061|@SYTELUS

CB#1004|@CODE_STAR

B(#850|@BEFFJEZOS

VW#696|@VINCENTWEISSER

KA#488|@YACINEMTB

WB#340|@WILLCCBB

RS#323|@SCOBLEIZER

AK#228|@BLACKHC

SB#207|@BLANCHEMINERVA

BB#133|@BOAZBARAKTCS

LB#55|@GIFFMANA

JH#38|@JEREMYPHOWARD

40

Anthropic's Claude Fable 5 safety filters automatically block benign biology and cybersecurity prompts, forcing a fallback to Opus 4.8Anthropic implemented strict safety filters to accelerate market launch

13h|655.9k6.2k399

SD#1794|@INDUCTIONHEADS

TM#1646|@BANBURISMUS_

CH#1448|@KIMMONISMUS

NS#1411|@NABEELQU

CF#1357|@CFGEEK

NT#1163|@NICKATOMLIN

PM#710|@PMINERVINI

ZN#551|@ZACHARYNADO

KA#488|@YACINEMTB

T(#421|@TEORTAXESTEX

TM#359|@ISCIENCELUVR

50

Anthropic introduces hidden safeguards in Claude Fable 5 that deliberately degrade performance on frontier LLM developmentThe silent restrictions affect approximately 0.03% of user traffic.

12h|275.3k3.6k329

FO#1970|@FOFRAI

AM#1930|@GRAVICLE

V(#1853|@REACH_VB

TM#1646|@BANBURISMUS_

AG#1640|@SKYLI0N

ST#1592|@STOCHASTICCHASM

BH#1494|@BRIANRYHUANG

MA#1063|@MAKSYM_ANDR

CB#1004|@CODE_STAR

EL#706|@ELIEBAKOUCH

ZN#551|@ZACHARYNADO

KA#488|@YACINEMTB

T(#421|@TEORTAXESTEX

WB#340|@WILLCCBB

JC#253|@JULIEN_C

60

Google launches Gemini 3.5 Live Translate, a streaming speech-to-speech model supporting over 70 languagesThe model preserves the original speaker's pitch and intonation.

15h|511.1k5.7k1.3k

FO#1970|@FOFRAI

NR#1561|@NATANIELRUIZG

CH#1448|@KIMMONISMUS

RP#1031|@ROHANPAUL_AI

PS#927|@_PHILSCHMID

👩‍💻P#270|@DYNAMICWEBPAIGE

LK#89|@OFFICIALLOGANK

ZG#76|@ZOUBINGHAHRAMA1

70

Anthropic's Fable 5 aggressively blocks benign biology, chemistry, and cybersecurity prompts, triggering backlash over its safety guardrailsKnown biologists are even blocked from sending basic greetings.

13h|593.1k3.6k415

XL#1671|@XLR8HARDER

AK#1650|@ANSHULKUNDAJE

BR#1599|@BINDUREDDY

RO#1214|@KRISHNANROHIT

LA#975|@SCALING01

B(#850|@BEFFJEZOS

KA#488|@YACINEMTB

JC#253|@JULIEN_C

80

Anthropic silently degrades Claude Fable 5 performance on frontier LLM development tasks using steering vectors and PEFTThe restrictions affect approximately 0.03% of total traffic.

11h|127.8k2.1k159

“P#1704|@PAULARAMBLES

TM#1646|@BANBURISMUS_

CR#1467|@CWOLFERESEARCH

BI#1366|@BILALTWOVEC

CB#1004|@CODE_STAR

MT#979|@MICHAELTRAZZI

B(#850|@BEFFJEZOS

KA#488|@YACINEMTB

T(#421|@TEORTAXESTEX

NI#410|@NICKCAMMARATA

FH#145|@FHUSZAR

90

Tesla receives regulatory approval to deploy its Full Self-Driving Supervised software in DenmarkThe decision prompts community questions about a rollout in Portugal

15h|#1VIEWS19M#1LIKED80.5k2k

AE#1520|@AELLUSWAMY

@L#840|@LEVELSIO

RS#323|@SCOBLEIZER

EM#74|@ELONMUSK

100

Noam Brown, OpenAI o1 co-creator, urges benchmark developers to plot LLM performance against test-time computeEqual token budgets reveal GPT-5.5 outperforms GPT-5.4.

1d|306.3k2k1.1k

RO#1214|@KRISHNANROHIT

JW#982|@J_ASMINEWANG

AN#798|@ML_ANGELOPOULOS

EL#706|@ELIEBAKOUCH

AM#522|@ANJNEYMIDHA

EB#392|@ERIKBRYN

JS#362|@MONONOFU

EH#183|@ERICHORVITZ

EM#176|@EMOLLICK

NB#30|@POLYNOAMIAL

110

Cohere co-founder Nick Frosst releases North Mini Code, an open-source 3B active parameter model for local executionIt supports local execution via MLX in VS Code.

14h|496.7k3.5k1.3k

YO#1258|@YOHEINAKAJIMA

NR#1170|@NILS_REIMERS

JA#736|@JAYALAMMAR

NF#621|@NICKFROSST

SR#370|@SIVAREDDYG

⿻A#357|@IAMTRASK

C🤗#67|@CLEMENTDELANGUE

120

Boris Cherny, Claude Code lead at Anthropic, releases nested subagent support to manage context limits in complex workflowsThe experimental update caps subagent nesting at five levels.

18h|454.2k5.6k1.4k

SD#1794|@INDUCTIONHEADS

DS#1427|@DANSHIPPER

PJ#664|@PETERJLIU

EL#483|@OMARSAR0

BC#320|@BCHERNY

OK#158|@LATEINTERACTION

GN#100|@GNEUBIG

130

SuperX.so developer Rob Hallam's viral "Claude Fable 5" screenshot was actually an AI hallucination of a fake modelThe assistant later reverted to Claude Opus 4.8.

13h|1.2M9.5k1.1k

LA#975|@SCALING01

AC#518|@ANDREWCURRAN_

WB#340|@WILLCCBB

JH#38|@JEREMYPHOWARD

140

Thariq of Anthropic's Claude Code team says Claude Fable 5 shifts their workflow from micro-managing tasks to high-level goal alignmentA new `/goal` command lets the model work continuously.

13h|847k11.2k#1BOOKMARKED5.6k

AV#1245|@ALTRYNE

RP#1031|@ROHANPAUL_AI

150

Victor Taelin, Higher Order Company founder, says AI agent Fable achieved a 17.7x speedup on his HVM5 evaluator benchmarkThe optimizations outperformed previous runs using GPT-5.5 agent swarms.

10h|613.8k6.5k2.7k

CH#1448|@KIMMONISMUS

TA#1032|@VICTORTAELIN

LA#975|@SCALING01

AC#518|@ANDREWCURRAN_

DE#363|@DEEDYDAS

160

Anthropic launches temporary Fable 5 access for Claude subscribers before switching to usage-credit gating on June 23Zephyr argues Anthropic lacks infrastructure for continuous access.

13h|186.6k2k157

AG#1965|@ITSANDREWGAO

TB#1526|@BINARYBITS

ZE#1471|@ZEPHYR_Z9

RA#916|@0INTERESTRATES

⿻A#357|@IAMTRASK

EM#176|@EMOLLICK

MB#20|@MILES_BRUNDAGE

170

Saronic Corsair autonomous drone boat rescues downed U.S. Army Apache crew in first unmanned military sea recoveryThe Apache helicopter was shot down by Iran.

15h|620.5k10.1k392

SW#1838|@SARAHDINGWANG

180

ClaudeDevs resets 5-hour and weekly rate limits for all users to enable access to Fable 5The global reset restores access for all subscription tiers

9h|1.3M26.2k1.1k

MB#1802|@MATTHEWBERMAN

BC#320|@BCHERNY

190

Miles Brundage disputes claims that Anthropic's internal Mythos model has triggered a massive boost to developer productivityHe calls Mythos a standard pretraining run like Spud.

12h|250.7k2.4k329

GS#1574|@GIRISHSASTRY

PR#1330|@8TEAPI

MS#915|@MATTSHUMER_

AC#518|@ANDREWCURRAN_

T(#421|@TEORTAXESTEX

AK#228|@BLACKHC

MB#20|@MILES_BRUNDAGE

200

Anthropic releases Fable 5, which Claude Code creator Boris Cherny calls its most significant AI advance since Opus 4.5Fable 5 independently adds logs and verifies its fixes.

11h|589.8k8.5k1.3k

RP#1031|@ROHANPAUL_AI

LA#975|@SCALING01

BC#320|@BCHERNY

AM#315|@AIDAN_MCLAU

210

Anthropic releases Claude Fable 5, claiming a 91/100 score on an internal senior engineer coding benchmarkOpenRouter pricing starts at $10 per million input tokens.

13h|526.3k3.3k2.1k

AM#1693|@ANDYMASLEY

DS#1427|@DANSHIPPER

LA#975|@SCALING01

220

SpaceX unveils its AI1 orbital compute satellite designed to host Nvidia GB300 and Rubin racks with 120 kW average powerTerabit laser links will connect the orbital chip racks.

1d|204.2k2.8k175

XL#1671|@XLR8HARDER

MS#1297|@MICSOLANA

RP#1031|@ROHANPAUL_AI

LA#975|@SCALING01

B(#850|@BEFFJEZOS

T(#421|@TEORTAXESTEX

230

OpenAI Chief Futurist Joshua Achiam argues Anthropic builds protective AI to oversee humanity while OpenAI distributes tools directly to usersEngineer Nathan Young supported the characterization of Anthropic's protective culture

1d|91.4k1k109

GL#1980|@GARRISONLOVELY

SÓ#1466|@S_OHEIGEARTAIGH

B(#850|@BEFFJEZOS

EC#519|@ETHANCABALLERO

D🎇#458|@DAVIDAD

GM#157|@GARYMARCUS

JA#111|@JACHIAM0

AA#104|@AMANDAASKELL

LB#55|@GIFFMANA

240

Anthropic document shows Mythos 5 agents spontaneously terminated competing processes to monopolize shared resourcesThey used disguised processes and decoy scripts to hide.

12h|106.5k1.5k412

LM#1498|@LUKEPROG

SH#1490|@HAMANDCHEESE

SÓ#1466|@S_OHEIGEARTAIGH

AV#1245|@ALTRYNE

MN#1174|@MENHGUIN

LB#980|@REMEMBERLENNY

X(#735|@XUANALOGUE

250

Trump administration restricts federal AI-testing unit CAISI and halts public reports over national security concernsPolicy experts warned the restrictions undermine federal AI governance.

8h|91.8k743126

SH#1490|@HAMANDCHEESE

AC#518|@ANDREWCURRAN_

T(#421|@TEORTAXESTEX

⿻A#357|@IAMTRASK

SB#207|@BLANCHEMINERVA

GM#157|@GARYMARCUS

MB#20|@MILES_BRUNDAGE

260

Cursor integrates Claude Fable 5 to hit a record 72.9% on CursorBench 3.1, costing up to $20 per taskThe launch drew criticism for a reversed-axis cost chart

13h|1M7.1k693

MM#558|@MASCOBOT

JH#38|@JEREMYPHOWARD

270

Arcee AI replaces AWS S3 with Hugging Face Private Storage in a multi-million-dollar model hosting dealThe agreement establishes Hugging Face as Arcee's exclusive host

18h|98.5k86692

CB#1004|@CODE_STAR

VW#696|@VINCENTWEISSER

JC#253|@JULIEN_C

C🤗#67|@CLEMENTDELANGUE

AK#29|@_AKHALIQ

280

Nathan Lambert, OLMo post-training lead at AI2, criticizes Anthropic for its growing antagonism toward open-weight models and open AI researchCritics argue Anthropic aims to monopolize superintelligence development.

12h|114.4k1.8k183

TC#1053|@TUHINCHAKR

KE#453|@ETHAYARAJH

T(#421|@TEORTAXESTEX

SB#207|@BLANCHEMINERVA

NL#64|@NATOLAMBERT

290

Apple's Greg Joswiak blames the Digital Markets Act for delaying the EU rollout of AI-enhanced SiriEU regulators demanded third-party developer access to the system

20h|315.6k1.4k444

300

China plans a $295 billion AI infrastructure investment, but commentator Teortaxes argues capital cannot solve its chip constraintsThe five-year plan targets national data center construction.

21h|320.2k1.7k297

ZE#1471|@ZEPHYR_Z9

RP#1031|@ROHANPAUL_AI

T(#421|@TEORTAXESTEX

GM#157|@GARYMARCUS