Tech News

Today's Highlights

In case you missed it

Ramp and Revelio Labs study of 21,559 firms finds high-intensity AI adopters grew headcount by 10% over two years

SI#584|@SIGNULLL

NB#353|@NATHANBENAICH

It’s a new day at Digg. Fresh stories are clustering now. 15% complete.

Top Stories

Posts:—|Clusters:—|Next Crawl:—

10

Anthropic pulls Claude Fable 5 offline one day after redeployment following a successful jailbreak by security researcher PlinyThe model was redeployed following discussions with the US government.

1h|3.4k200

CH#1364|@KIMMONISMUS

LA#1218|@SCALING01

B(#530|@BEFFJEZOS

Top post: @beffjezos “Ok thank God”

20

Google DeepMind's Pranav Shyam argues short-horizon RL tasks act like bandit problems, making value functions unnecessary and adding PPO latencyGrad says RL trainers aren't bottlenecks at high token scales.

21m|2422

PS#1597|@RECURSEPARADOX

GR#1525|@GRAD62304977

Top post: @recurseparadox “@Grad62304977 Basically advantage functions reduce variance and longer chains have high variance. For low variance setups GRPO sampling and pool is good enough signal”

30

Elie Bakouch says Claude's safety filters blocked a routine request to retrieve an Artifact link from a past sessionThe filter flagged the cross-session request as a policy violation

1h|10020

FB#1782|@XEOPHON

EL#1138|@ELIEBAKOUCH

Top post: @eliebakouch “@xeophon you know the worst part, the color was like blue for claude and orange for codex and just asked in another convo to switch it and send the artifact link 😭”

Watchlist

Video Signals

FO#1857|@FOFRAI

AI Subagent Scripts 11 Hand Transformations Using Omni and Lyria 3

40m ago|Views 850Likes 10Bookmarks 4

RP#1260|@ROHANPAUL_AI

Perplexity CEO: Power Users Now Consume Millions in AI Compute

1h ago|Views 973Likes 0Bookmarks 0

SA#1893|@SVPINO

SuperNori Debuts Proactive AI Agent For Family Caretakers

1h ago|Views 2.2KLikes 0Bookmarks 0

Recent Stars

karpathy/nanochat3H AGO
Trains minimal hackable GPT-style LLMs end-to-end from tokenization through pretraining, finetuning, eval, inference and chat UI on one GPU node.
278
55.7k stars
p-org/P3H AGO
Models distributed systems as communicating state machines with automated formal verification for correctness.
278
3.6k stars
…ab/gram-newton-schulz3H AGO
Implements Gram Newton-Schulz iteration on Gram matrices for faster polar decomposition in Muon.
1070
161 stars
jotaen/kong-completion5H AGO
Provides tab completions for CLI apps built with the kong parser in bash, zsh and fish.
382
19 stars
…sariumut/FaceAnything5H AGO
Reconstructs high-fidelity 4D faces and dense tracks from arbitrary image sequences via canonical facial coordinate prediction.
484
53 stars

Github Stars

(7 days)

xlang-ai/OSWorld-V22D AGO
Benchmarks computer-use agents on long-horizon real-world desktop tasks via VM environments and gated task sets.
1015
397
844
113 stars
deepseek-ai/DeepSpec3D AGO
Provides full-stack code for training/evaluating speculative decoding draft models with data prep and eval scripts.
484
1197
1138
2.5k stars
THUDM/slime4D AGO
Integrates Megatron training with SGLang rollouts for LLM post-training and RL scaling.
458
1015
484
7k stars
…nd-models-of-ai-tools5D AGO
Aggregates leaked and open-sourced system prompts plus tool definitions from AI coding agents like Cursor, Claude, and Devin.
1015
827
141.3k stars
anthropics/skills5D AGO
Hosts collections of self-contained skills, specs, and templates for dynamically extending Claude's capabilities on specialized tasks.
1015
1470
155.8k stars

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

Yesterday's Top Stories, Jun 30, 2026.

Posts: 4,690|Clusters: 1,451|Stories frozen at midnight PT

10

Ramp and Revelio Labs study of 21,559 firms finds high-intensity AI adopters grew headcount by 10% over two yearsEntry-level headcount at those high-adopting firms rose 12%.

19h|318.8k2.2k694

DJ#1711|@DAN_JEFFRIES1

CB#1678|@CODE_STAR

TN#1159|@TURNERNOVAK

CV#659|@C_VALENZUELAB

SI#584|@SIGNULLL

NB#353|@NATHANBENAICH

20

Anthropic will restore access to Claude Fable 5 and Mythos 5 after US lifts export controlsService restoration begins July 2 following policy negotiations

8h|000

“P#1850|@PAULARAMBLES

MB#1784|@MATTHEWBERMAN

BH#1766|@BENHYLAK

GI#1735|@GREGISENBERG

ZE#1700|@ZEPHYR_Z9

ND#1657|@NICKADOBOS

TH#1481|@TIMHWANG

CB#1432|@CHRISJBAKKE

AV#1423|@ALTRYNE

CH#1364|@KIMMONISMUS

RO#1319|@KRISHNANROHIT

RP#1260|@ROHANPAUL_AI

MI#1231|@MARIYAIVASILEVA

SM#1228|@PITDESI

LA#1218|@SCALING01

BA#1116|@BAYESLORD

YO#1082|@YOHEINAKAJIMA

🎭#1015|@DEEPFATES

DS#976|@DANSHIPPER

LG#938|@LOGANGRAHAM

JL#907|@JERRYJLIU0

KD#905|@CSPROFKGD

RS#741|@ZIV_RAVID

MW#717|@MENGDIWANG10

MM#716|@MASCOBOT

HE#708|@HEINRICHKUTTLER

MS#691|@MATTSHUMER_

AC#682|@ANDREWCURRAN_

J⧉#674|@REPLIGATE

DA#619|@GROWING_DANIEL

TM#613|@ISCIENCELUVR

YJ#586|@YUCHENJ_UW

B(#530|@BEFFJEZOS

T(#501|@TEORTAXESTEX

RS#447|@SCOBLEIZER

T🪽#331|@TEKNIUM

MB#57|@MILES_BRUNDAGE

30

Meituan releases LongCat-2.0, a 1.6-trillion-parameter open-weights MoE model trained on 50,000 Chinese ASICsIt supports a native 1 million token context window

1d|96.1k1.3k198

SO#1923|@CIS_FEMALE

BR#1865|@BRIANROEMMELE

LH#1589|@OHLENNART

RP#1260|@ROHANPAUL_AI

CP#856|@CHRIS_J_PAXTON

T(#501|@TEORTAXESTEX

RS#447|@SCOBLEIZER

JC#335|@JULIEN_C

C🤗#109|@CLEMENTDELANGUE

NL#80|@NATOLAMBERT

OP#78|@OFIRPRESS

40

Etched raises $800 million to ship its custom AI inference hardware starting this summerThe company has secured over $1 billion in contracts

15h|000

ID#1975|@IAINDUNNING

SO#1923|@CIS_FEMALE

BR#1865|@BRIANROEMMELE

AG#1852|@ITSANDREWGAO

MP#1823|@MVPATEL2000

DZ#1798|@DZHANG50

LA#1787|@LIORONAI

NA#1765|@NIKESHARORA

BT#1608|@BENTOSSELL

MB#1592|@BAKKERMICHIEL

SW#1519|@0XSIGIL

LP#1446|@LPOLOVETS

CH#1364|@KIMMONISMUS

JS#1317|@JOON_S_PK

MO#1229|@MYLEOTT

LA#1218|@SCALING01

JH#1193|@MATHEMAGIC1AN

YP#1173|@YPATIL125

EL#1138|@ELIEBAKOUCH

SH#1086|@SONYATWEETYBIRD

TS#1051|@TIMSWEENEYEPIC

DR#982|@DANINTHEORY

"M#867|@LM_BRASWELL

DF#756|@DANIELLEFONG

DG#718|@DEMI_GUO_

MS#691|@MATTSHUMER_

SB#664|@SCOTTBELSKY

GA#598|@GABRIEL1

CC#594|@ITSCLIVETIME

PO#571|@PATRICK_OSHAG

SW#548|@SCOTTWU46

B(#530|@BEFFJEZOS

WM#520|@WILLMANIDIS

AC#484|@ANDREW_N_CARR

AM#437|@ANJNEYMIDHA

BJ#316|@BRYAN_JOHNSON

JT#138|@MILLIONINT

SE#125|@STEFANOERMON

TD#112|@TRI_DAO

RA#102|@_AROHAN_

50

Rihard Jarc argues token optimization structurally benefits Amazon, Microsoft, and Google over dedicated AI labsBill Ackman cited the analysis to back his thesis.

20h|#1VIEWS1M#1LIKED3.8k#1BOOKMARKED2.8k

DA#1472|@DAVIDLEE

C🤗#109|@CLEMENTDELANGUE

60

Spotted backend preparations spark speculation that a Claude Sonnet 5 release is imminentThe launch may coincide with a rumored Fable 5.

21h|109.1k90269

CH#1364|@KIMMONISMUS

RS#741|@ZIV_RAVID

B(#530|@BEFFJEZOS

70

Perplexity founder Aravind Srinivas says US chip export controls accelerated China's physical AI infrastructure buildoutChina builds data centers faster without power or regulatory constraints.

23h|154.6k611237

RP#1260|@ROHANPAUL_AI

80

Anthropic redeploys Claude Fable 5 globally after updating safety classifiers to block a vulnerability exploit in 99% of casesCoding and debugging tasks temporarily route to Opus 4.8.

3h|000

PC#1909|@PHILHCHEN

BR#1865|@BRIANROEMMELE

XL#1859|@XLR8HARDER

AG#1852|@ITSANDREWGAO

OM#1837|@OMOORETWEETS

AS#1813|@ALEXSTAMOS

BH#1766|@BENHYLAK

BS#1760|@BILAWALSIDHU

ZE#1700|@ZEPHYR_Z9

ND#1657|@NICKADOBOS

CH#1364|@KIMMONISMUS

JL#907|@JERRYJLIU0

PT#821|@ELDER_PLINIUS

MM#716|@MASCOBOT

MS#691|@MATTSHUMER_

AC#682|@ANDREWCURRAN_

DA#619|@GROWING_DANIEL

B(#530|@BEFFJEZOS

RS#447|@SCOBLEIZER

SB#289|@BLANCHEMINERVA

MB#57|@MILES_BRUNDAGE

90

20VC's Harry Stebbings reports five founders claim they cut AI inference costs by 75% with minimal effortThe companies ranged from startups to a $200B corporation.

19h|131.5k714269

HS#800|@HARRYSTEBBINGS

100

Khosla Ventures' Keith Rabois advocates using AI diagnostic tools directly instead of relying solely on traditional doctorsHe cites OpenAI's o3 successfully diagnosing rare pediatric diseases.

19h|276.9k510445

DF#1647|@DAVEFONTENOT

110

Anthropic launches Claude Sonnet 5 with a 1M context, scoring 63.2% on SWE-bench ProTesting revealed the model occasionally produces unprompted cosmic prayers.

17h|000

BR#1865|@BRIANROEMMELE

XL#1859|@XLR8HARDER

FB#1782|@XEOPHON

PS#1581|@SKIRANO

AV#1423|@ALTRYNE

CH#1364|@KIMMONISMUS

RO#1319|@KRISHNANROHIT

RP#1260|@ROHANPAUL_AI

LA#1218|@SCALING01

TA#1142|@VICTORTAELIN

RS#741|@ZIV_RAVID

EL#684|@OMARSAR0

AC#682|@ANDREWCURRAN_

J⧉#674|@REPLIGATE

TM#613|@ISCIENCELUVR

WB#573|@WILLCCBB

T(#501|@TEORTAXESTEX

RS#447|@SCOBLEIZER

KR

Digg Deeper question: “Compare this models SWE-bench Pro vs all other frontier models side-by-side, just give me models and scores”

answerview answer

120

Discovered Claude app strings show Anthropic plans to gate Fable 5 behind metered credits and identity verificationThe identity verification requirement contradicts Anthropic's previous statements

1d|135k847116

CH#1364|@KIMMONISMUS

130

Guardian profile of Google DeepMind's Iason Gabriel questions if corporate ethicists can steer frontier AI developmentGabriel joined the lab in 2017 as an early ethicist.

22h|32.3k18098

IG#914|@IASONGABRIEL

SK#667|@SEBKRIER

RL#547|@RYAN_T_LOWE

140

DeepSeek releases DSpark, a speculative decoding algorithm boosting DeepSeek-V4 throughput by up to 661%Open-source vLLM has rapidly integrated support for the algorithm.

22h|70.1k41966

ZE#1700|@ZEPHYR_Z9

T(#501|@TEORTAXESTEX

KA#403|@YACINEMTB

AD

Digg Deeper question: “What's speculative decoding? And why does this news matter?”

answerview answer

150

Google DeepMind opens its new London headquarters, named Platform 37 after AlphaGo's historic Move 37DeepMind staff shared first-day tours of the terraced workspace.

21h|42.9k25423

FO#1857|@FOFRAI

PS#1408|@_PHILSCHMID

JP#774|@JPARKERHOLDER

PV#647|@PETARV_93

TK#291|@TEJASDKULKARNI

AD

Digg Deeper question: “Why are AI companies opening offices in London? Is there a recent event that happened to cause this or is this part of a broader trend?”

answerview answer

160

Alibaba's Qwen team disputes researcher Tianhang Zhu's claims of leading its reinforcement learning developmentZhu's Google Scholar profile listed papers lacking his name.

23h|48.1k42043

T(#501|@TEORTAXESTEX

SZ#84|@SUCHENZANG

170

Anthropic's Claude Code covertly fingerprints Chinese users using XOR-obfuscated timezone checks and hidden prompt telemetryThe covert tracking began in Claude Code version 2.1.91.

19h|7k4611

BR#1865|@BRIANROEMMELE

ZE#1700|@ZEPHYR_Z9

VI#1634|@VIKHYATK

RP#1260|@ROHANPAUL_AI

T(#501|@TEORTAXESTEX

T🪽#331|@TEKNIUM

180

Open-source contributor Grad says applying length penalties to software engineering tasks first prevents redundant reinforcement learning trainingThe ablation studies were performed on Qwen3-30B-A3B.

20h|7.1k5636

GR#1525|@GRAD62304977

T(#501|@TEORTAXESTEX

190

Instructor creator Jason Liu floats hypothetical OpenAI arts grants, prompting proposals to fund artists pushing prompting limitsOpenAI has not announced any official arts grant program.

22h|24.3k26935

200

ETH Zürich's Florian Tramèr questions ACL 2026 conference budget allocating $50 per coffee break with no actual mealsOrganizers claim food costs will exceed $800 per attendee.

20h|15.9k649

FT#1663|@FLORIAN_TRAMER

KD#905|@CSPROFKGD

GK#197|@THEGAUTAMKAMATH

YA#169|@YOAVARTZI

210

Conception reports generating the first early-stage human eggs derived from stem cellsThe process uses iPSCs to generate early-stage primordial follicles

15h|000

TH#1985|@VOOOOOOGEL

KJ#1974|@KATIES

M🍥#1959|@MAYFER

JH#1658|@JOHNOHALLMAN

MD#1632|@MHDEMPSEY

MP#1029|@EASTDAKOTA

DF#756|@DANIELLEFONG

BA#398|@BRIAN_ARMSTRONG

BP#381|@BORISMPOWER

JA#173|@JACHIAM0

220

Spotify VP Niklas Gustavsson says AI assists 73% of pull requests, sparking debate over platform reliabilityCritics cited three recent platform outages over four weeks

1d|24.4k29323

VI#1634|@VIKHYATK

RS#741|@ZIV_RAVID

230

OpenAI internal optimization cuts inference costs by half, running logged-out ChatGPT traffic on a couple hundred GPUsObservers speculate the undisclosed method relies on speculative decoding

15h|000

DM#1770|@DAVEMCCLURE

CB#1678|@CODE_STAR

CH#1364|@KIMMONISMUS

RP#1260|@ROHANPAUL_AI

LA#1218|@SCALING01

JL#1087|@JESSICALESSIN

AC#682|@ANDREWCURRAN_

⿻A#595|@IAMTRASK

T(#501|@TEORTAXESTEX

NL#80|@NATOLAMBERT

240

Google DeepMind launches Nano Banana 2 Lite for $0.034 images and previews Gemini Omni Flash APIOmni Flash generates 10-second videos at $0.10 per second.

14h|000

FO#1857|@FOFRAI

AV#1423|@ALTRYNE

PS#1408|@_PHILSCHMID

RP#1260|@ROHANPAUL_AI

EL#684|@OMARSAR0

T(#501|@TEORTAXESTEX

👩‍💻P#387|@DYNAMICWEBPAIGE

PM#317|@DOCMILANFAR

JB#245|@JASONBALDRIDGE

LK#100|@OFFICIALLOGANK

DH#29|@DEMISHASSABIS

250

Visa, Stripe, and Coinbase back Open Standard's upcoming OpenUSD stablecoin designed to share economics with usersTempo will natively issue the stablecoin on day one.

17h|000

DM#1134|@DAVIDMARCUS

MH#1109|@MATTHUANG

MP#1029|@EASTDAKOTA

AL#820|@ALFRED_LIN

BA#398|@BRIAN_ARMSTRONG

EM#171|@EMOSTAQUE

PC#81|@PATRICKC

AD

Digg Deeper question: “How will this affect popular stablecoins like USDC and USDT?”

answerview answer

260

Smol.ai founder Shawn 'swyx' Wang kicks off the AI Engineer World's Fair, celebrating three years of the developer summitThe developer summit features a keynote by creator Theo.

17h|000

V(#1942|@REACH_VB

“P#1850|@PAULARAMBLES

AV#1423|@ALTRYNE

AE#1420|@EMBIRICO

AM#1092|@ANDREWDAI

RH#642|@ROMAINHUET

RS#447|@SCOBLEIZER

PS#382|@STEIPETE

270

UBTech launches its U1 bionic robots in China starting at 119,800 RMB with a local emotional AI modelThe local AI is trained on Huawei's Ascend framework

19h|11.6k12813

CP#856|@CHRIS_J_PAXTON

T(#501|@TEORTAXESTEX

280

AI developer Teortaxes says Meituan’s 50,000-chip cluster actually uses 25,000 physical dual-die Huawei Ascend 910C processorsThe TSMC-manufactured processors were obtained via the Sophgo diversion.

21h|4.2k3313

LH#1589|@OHLENNART

T(#501|@TEORTAXESTEX

290

Overheard anecdote: CMU's Niloofar Mireshghallah compares the risk of AGI obsoleting universities to building a wrapper on ClaudeAn interlocutor claimed AGI will make academics a permanent underclass.

15h|000

HH#1686|@HAMELHUSAIN

TC#967|@TUHINCHAKR

SS#518|@SH_REYA

KC#428|@KAMALIKAC

N✈️#287|@NILOOFAR_MIRE

GD#223|@GREGD_NLP

300

Sebastian Raschka releases a new book on building reasoning LLMs from scratchThe guide covers inference scaling, reinforcement learning, and distillation.

17h|000

MJ#1437|@MIMANSAJ

EL#1138|@ELIEBAKOUCH

CA#1060|@CHRISALBON

KM#409|@KORYMATH

NL#80|@NATOLAMBERT