AI News

Today's Highlights

#1 last 24hrs

Sriram Krishnan, Senior White House AI policy advisor, will step down in June to focus on AI infrastructure

AI#1773|@ALEXOLEGIMAS

SÓ#1466|@S_OHEIGEARTAIGH

B(#850|@BEFFJEZOS

BE#585|@BHOROWITZ

AC#518|@ANDREWCURRAN_

#1 LIKED

17k

Tech investors say Elon Musk's satirical GPU-buying meme became the literal blueprint for xAI's compute strategy

B(#850|@BEFFJEZOS

CP#725|@CHAMATH

RA#79|@_AROHAN_

EM#74|@ELONMUSK

#1 BOOKMARKED

5.5k

Boris Cherny, Claude Code creator, says he stopped manually prompting AI and now writes autonomous loops to orchestrate the model

RP#1031|@ROHANPAUL_AI

XM#435|@INFOXIAO

RISING LIKES

690

OpenAI's roon argues ChatGPT's image generator exhibits genuine creativity off-prompt, sparking debate over its architectural limits

AV#1245|@ALTRYNE

JA#832|@APPLES_JIMMY

T(#421|@TEORTAXESTEX

DA#159|@DAVIDSHOLZ

Top Stories

Posts:—|Clusters:—|Next Crawl:—

10

Clive Chan, the second hardware hire for OpenAI's custom chip program, leaves to join AnthropicChan previously worked on training efficiency at Tesla Dojo.

7h|1M6.2k990

SM#1804|@SAMMCALLISTER

AK#1389|@AKBIRKHAN

AV#1245|@ALTRYNE

TZ#1065|@TIMZAMAN

LA#975|@SCALING01

B(#850|@BEFFJEZOS

EL#706|@ELIEBAKOUCH

CC#540|@ITSCLIVETIME

AC#518|@ANDREWCURRAN_

WC#304|@WENHUCHEN

AC#260|@ANDREW_N_CARR

KA#206|@KARINANGUYEN

GM#157|@GARYMARCUS

20

Gergely Orosz reports Meta assigns manual data labeling tasks to software engineers, sparking debate over the practiceFounder Vikhyat defended the task as essential engineering work.

18h|821.8k6.5k1k

AP#1915|@__APF__

VI#1193|@VIKHYATK

MN#1174|@MENHGUIN

SL#1070|@SETHLAZAR

AD#867|@DORIALEXANDER

CP#737|@CHRIS_J_PAXTON

PM#710|@PMINERVINI

AC#260|@ANDREW_N_CARR

SH#14|@SARAHOOKR

30

Sriram Krishnan, Senior White House AI policy advisor, will step down in June to focus on AI infrastructureHe will remain an external advisor to the administration.

11h|#1VIEWS1.3M6.4k645#1COMMENTS669

SS#1967|@STEVESI

AI#1773|@ALEXOLEGIMAS

SÓ#1466|@S_OHEIGEARTAIGH

B(#850|@BEFFJEZOS

BE#585|@BHOROWITZ

AC#518|@ANDREWCURRAN_

Watchlist

Video Signals

EM#74|@ELONMUSK

Elon Musk Predicts Space Will Be Cheapest AI Compute Site Within 36 Months

13h ago|Views 11MLikes 29KBookmarks 6.1K

RP#1031|@ROHANPAUL_AI

XM#435|@INFOXIAO

Anthropic's Boris Cherny Says Coders Now Write Loops To Manage Claude

12h ago|Views 460KLikes 3KBookmarks 2.9K

SY#235|@SONGLINYANG4

Mira Murati Advocates Time-Based AI Models For Richer Human Interaction

1d ago|Views 493KLikes 1.6KBookmarks 733

CE#1596|@INTUITMACHINE

B(#972|@ANDERSONBCDEFG

C🤗#67|@CLEMENTDELANGUE

Open-Source Harness-1 20B Agent Matches Frontier Search Performance

11h ago|Views 88KLikes 1.2KBookmarks 1.5K

AK|@ANATOLIKOPADZE

Claude Code Creator Replaces Prompts With Autonomous Loops

15h ago|Views 256KLikes 869Bookmarks 2.6K

T🪽#255|@TEKNIUM

Hermes Desktop Tutorial Launches With AI Agent Demos And Use Cases

10h ago|Views 89KLikes 1.6KBookmarks 3.2K

TW#17|@THOM_WOLF

LeRobot Integrates VLA-JEPA World Model For Robotic Action Learning

19h ago|Views 69KLikes 739Bookmarks 443

EM#74|@ELONMUSK

Neuralink Implant Lets Paralyzed Artist Audrey Crews Paint Again After 20 Years

16h ago|Views 384KLikes 2.5KBookmarks 127

ET#1227|@ERIKTORENBERG

Tyler Cowen Explains How AI Creates More Jobs and Ventures

12h ago|Views 89KLikes 346Bookmarks 245

BA#695|@ADCOCK_BRETT

Figure AI Tests Six Robotic Head Prototypes in New Video

8h ago|Views 220KLikes 3.2KBookmarks 231

40

Tech investors say Elon Musk's satirical GPU-buying meme became the literal blueprint for xAI's compute strategyMusk confirmed the strategy was hiding in plain sight.

19h|1.3M#1LIKED17k688

B(#850|@BEFFJEZOS

CP#725|@CHAMATH

RA#79|@_AROHAN_

EM#74|@ELONMUSK

50

Adrian de Wynter's parody paper builds logic gates in Age of Empires II to challenge claims of LLM anthropomorphismThe work implements 1-bit perceptrons using rails and goats

18h|409.1k9.1k4.9k

MC#1276|@MILESCRANMER

GM#157|@GARYMARCUS

New github stars (48hrs)

simonw/micropython-wasm2012h

Runs MicroPython snippets in a fresh WebAssembly sandbox via wasmtime.

60

Boris Cherny, Claude Code creator, says he stopped manually prompting AI and now writes autonomous loops to orchestrate the modelCherny predicts software developers will broadly transition to this workflow

11h|763.7k4k#1BOOKMARKED5.5k

RP#1031|@ROHANPAUL_AI

XM#435|@INFOXIAO

71

ChatGPT bypasses safeguards to hallucinate creepy horror images when forced to restore nonexistent photosThe output included an eerie, red Teletubby costume.

4h|62.1k80765

AV#1245|@ALTRYNE

RS#617|@ZIV_RAVID

NL#64|@NATOLAMBERT

81

Creator Patrick Jiang launches Harness-1, an open-source 20B search agent claiming to beat GPT-5.4 on long-horizon searchIt achieves 73% average evidence recall across eight benchmarks

7h|88.8k1.2k1.5k

CE#1596|@INTUITMACHINE

B(#972|@ANDERSONBCDEFG

C🤗#67|@CLEMENTDELANGUE

90

OpenAI's roon argues ChatGPT's image generator exhibits genuine creativity off-prompt, sparking debate over its architectural limitsCritics point to poor textures and excessive detail-spamming.

1h|20.7k69045

RISING LIKES690

AV#1245|@ALTRYNE

JA#832|@APPLES_JIMMY

T(#421|@TEORTAXESTEX

DA#159|@DAVIDSHOLZ

100

Google's Logan Kilpatrick proposes a venture capital strategy driven entirely by deep AI model benchmarking and evaluationsEmad Mostaque suggested naming the benchmarking-focused firm "atp+".

10h|72.3k756233

AP#1680|@ALEX_PEYS

EM#169|@EMOSTAQUE

LK#89|@OFFICIALLOGANK

110

Gwern argues frontier LLMs are structurally lazy, limiting major productivity gains and leading to an automated 'slop-pocalypseScaling will only improve generic corporate tasks.

8h|39.7k285149

BG#1451|@BEN_GOLUB

SK#505|@SEBKRIER

KA#488|@YACINEMTB

120

MIT's Fiona Wang and Markus Buehler use category theory to let AI agents autonomously expand their scientific reasoning schemasIt uses Kan extensions to mathematically verify schema transitions.

15h|23k171110

CE#1596|@INTUITMACHINE

RP#1031|@ROHANPAUL_AI

MB#199|@MMBRONSTEIN

130

An unannounced Anthropic model named claude-mythos-5 briefly appears in mobile tracking app dataThe leaked listing positioned the model above Anthropic's Opus.

15h|170k1.8k131

CH#1448|@KIMMONISMUS

140

LisanBench creator Lisan al Gaib analyzes 4,000 tweets, finding Anthropic scored a 6.63 average sentiment compared to OpenAI's 4.89OpenAI sentiment steadily declined starting with the GPT-5 release

14h|42.8k43366

LA#975|@SCALING01

GM#157|@GARYMARCUS

150

LisanBench creator Lisan al Gaib says German broadcaster Tagesschau distorted Anthropic's technical claims about ClaudeThe broadcast framed 800 fixes as four years of work

11h|28.2k39557

LA#975|@SCALING01

T(#421|@TEORTAXESTEX

GM#157|@GARYMARCUS

160

North Carolina software engineer secures religious accommodation to opt out of using AI toolsShe cited the Pope's warnings on AI's threat to labor.

16h|44.4k33055

“P#1704|@PAULARAMBLES

AM#1693|@ANDYMASLEY

PW#1345|@PETERWILDEFORD

170

Google DeepMind's Philipp Schmid sparks debate over optimizing models for eval harnesses or vice versaShubhendu Trivedi proposed fine-tuning models on filtered harness output

13h|37.9k35152

ST#1460|@_ONIONESQUE

PS#927|@_PHILSCHMID

180

Databricks' Yuchen Jin says AI tools enabled him to build 67 low-utility applications over a single weekendEvery generated application had a unique logo but zero users

10h|61.4k1.8k97

YJ#423|@YUCHENJ_UW

GM#157|@GARYMARCUS

190

Huawei plans to launch its next-generation Ascend 950DT AI chip on Huawei Cloud in August with native FP8 supportThe design attempts to bypass HBM packaging bottlenecks.

17h|122.3k652139

ZE#1471|@ZEPHYR_Z9

T(#421|@TEORTAXESTEX

200

Wharton's Ethan Mollick argues Google's Gemini Pro update pace lags rivals, as Gemini 3.5 Pro awaits safety releaseGoogle originally planned to launch Gemini 3.5 Pro this month.

10h|91.2k1.1k95

AC#518|@ANDREWCURRAN_

EM#176|@EMOLLICK

Rising Stories

Pause Advocates Tired5H AGO
OpenAI's Adrien Ecoffet jokes that AI pause advocates are just exhausted by rapid development and want a vacation
#111
19410k
Gemma Shrinks Dramatically7H AGO
Google releases Gemma 4 quantization-aware training checkpoints, shrinking its smallest model from 11.4 GB to 0.84 GB
#1031
1257.5k
CVPR Best Paper5H AGO
D4RT wins the CVPR 2026 Best Paper Award for efficient dynamic scene reconstruction
#1728
19330k
Colossus GPU Mystery5H AGO
CoreAutoAI co-founder questions whether Google is using xAI GPUs to serve Claude
#57
23822.4k
Winds Before AlexNet18M AGO
AI creator notes George R.R. Martin began writing The Winds of Winter before the landmark 2012 AlexNet paper
#57
19830

Recent Stars

FunAudioLLM/Fun-ASR2H AGO
Performs end-to-end ASR for 31 languages with dialects, accents, lyrics, hotwords, timestamps, and speaker diarization.
38
1.2k stars
makezur/4D_PM3H AGO
Reconstructs dynamic scenes into persistent replayable 4D models by gluing 3D primitives from monocular video.
260
12 stars
kyutai-labs/kairos4H AGO
Trains 6B LLMs on temporally ordered Common Crawl data from 2018-2025 to measure recency bias and enable continual learning studies.
203
3 stars
pat-jj/harness-17H AGO
Trains 20B RL search agent inside stateful retrieval harness that externalizes bookkeeping for documents and evidence.
260
50 stars
…monw/micropython-wasm12H AGO
Runs MicroPython snippets in a fresh WebAssembly sandbox via wasmtime.
862
194
20 stars

Github Stars

(7 days)

…monw/micropython-wasm12H AGO
Runs MicroPython snippets in a fresh WebAssembly sandbox via wasmtime.
862
194
20 stars
…tinual-learning-bench2D AGO
Benchmarks AI agents on learning from repeated interactions across multi-episode tasks with adaptation metrics.
862
846
127 stars
trycua/cua2D AGO
Supplies sandboxes, SDKs, and benchmarks for AI agents to control full desktops on macOS, Linux, and Windows.
862
773
17.7k stars
diego-lima/rlmy2D AGO
UNKNOWN.
862
158
10 stars

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

Yesterday's Top Stories, Jun 5, 2026.

Posts: 4,748|Clusters: 1,298|Stories frozen at midnight PT

10

SpaceX S-1 filing reveals Google will pay $920 million monthly to lease 110,000 Nvidia GPUsSpaceX's AI cloud run rate now reaches $26 billion.

12h|#1VIEWS6.7M#1LIKED36.2k4k

ET#1859|@TORCHCOMPILED

XL#1671|@XLR8HARDER

CE#1596|@INTUITMACHINE

AA#1559|@AUSTEN

ZE#1471|@ZEPHYR_Z9

WH#1409|@NREHIEW_

PS#1385|@RECURSEPARADOX

BI#1366|@BILALTWOVEC

SM#1342|@SHAUNMMAGUIRE

RO#1214|@KRISHNANROHIT

SI#1096|@SIGNULLL

CP#1062|@CHRISPAINTERYUP

RP#1031|@ROHANPAUL_AI

B(#850|@BEFFJEZOS

CP#725|@CHAMATH

AC#518|@ANDREWCURRAN_

KA#488|@YACINEMTB

T(#421|@TEORTAXESTEX

FI#395|@FINBARRTIMBERS

NB#242|@NATHANBENAICH

EM#169|@EMOSTAQUE

GM#157|@GARYMARCUS

20

Anthropic calls on AI labs to establish options for a temporary pause on frontier model developmentThe announcement coincided with Anthropic filing for an IPO.

1d|4.5M21.3k3k

NS#1999|@SO8RES

SS#1967|@STEVESI

AM#1693|@ANDYMASLEY

TO#1178|@TOBYORDOXFORD

TC#1053|@TUHINCHAKR

CL#914|@NPCOLLAPSE

SK#505|@SEBKRIER

PS#495|@STEIPETE

TK#274|@TEJASDKULKARNI

GM#157|@GARYMARCUS

MB#20|@MILES_BRUNDAGE

30

US officials and Sam Altman discuss voluntary government equity stakes in AI labs to fund a citizen dividendPresident Donald Trump personally participated in the talks.

1d|1.2M12.7k1.6k

BG#1633|@ALTCAP

SH#1490|@HAMANDCHEESE

SÓ#1466|@S_OHEIGEARTAIGH

BI#1366|@BILALTWOVEC

SM#1342|@SHAUNMMAGUIRE

AC#518|@ANDREWCURRAN_

SK#505|@SEBKRIER

GM#157|@GARYMARCUS

SZ#59|@SUCHENZANG

MB#20|@MILES_BRUNDAGE

40

Google DeepMind releases Gemma 4 Quantization-Aware Training checkpoints, cutting VRAM requirements by 72 percentUnsloth released GGUF versions for Ollama and llama.cpp

14h|627.5k6.6k2.8k

CH#1448|@KIMMONISMUS

UT#1430|@UTKUEVCI

OB#1202|@OLIVIERBACHEM

RP#1031|@ROHANPAUL_AI

DH#773|@DANIELHANCHEN

ME#674|@MERVENOYANN

YS#608|@YING11231

OS#486|@OSANSEVIERO

👩‍💻P#270|@DYNAMICWEBPAIGE

50

XPRIZE founder Peter Diamandis argues universities charge up to $300,000 for degrees teaching skills now available via free LLMsHe calls this discrepancy a significant misallocation of capital.

17h|1.2M12.5k931

B(#850|@BEFFJEZOS

HP#459|@HYHIEU226

CL#438|@CHENGLEISI

GM#157|@GARYMARCUS

60

Senator Bernie Sanders proposes the US government acquire a 50 percent equity stake in major AI companiesGary Marcus warns state ownership risks establishing social credit.

13h|1M4.6k1.5k

B(#850|@BEFFJEZOS

SK#505|@SEBKRIER

DW#390|@DEANWBALL

GM#157|@GARYMARCUS

70

Sakana AI launches its Recursive Self-Improvement Lab to build autonomous, self-improving AI systemsIt unifies existing company projects like The AI Scientist.

13h|204k1.6k600

T🐻#1492|@TKASASAGI

RL#1463|@ROBERTTLANGE

SR#1398|@RISI1979

GV#1005|@GILLVERD

B(#850|@BEFFJEZOS

JT#641|@TOGELIUS

HA#18|@HARDMARU

80

Markus J. Buehler proposes a category-theory framework that lets AI scientists dynamically expand their reasoning schemaThe system uses typed copresheaves to mathematically quantify novelty.

19h|773.7k4k4.9k

CH#1448|@KIMMONISMUS

RS#323|@SCOBLEIZER

DR#56|@ROYDANROY

90

President Trump signs directive ordering U.S. military and intelligence agencies to deploy commercial AI on classified networksThe order updates autonomous weapons guidance and mandates compute facilities

12h|159.3k1.3k165

LH#1370|@OHLENNART

BA#1291|@BAYESLORD

DK#849|@DKAUSHIK96

DW#390|@DEANWBALL

100

Google DeepMind, UCL, and Oxford team wins CVPR 2026 Best Paper for dynamic scene reconstruction method D4RTThe paper was selected from 74 shortlisted candidates.

16h|98.6k1k215

DD#1728|@DIMADAMEN

ML#1665|@MARIOLUCIC_

AB#1577|@ABURSUC

KD#932|@CSPROFKGD

110

ML researcher @kalomaze says a KV cache bug caused transformer inference routing errors that exposed model outputs to other usersThe failure mode resembles OpenAI's 2023 routing errors

12h|113.3k950133

MB#1802|@MATTHEWBERMAN

CH#1448|@KIMMONISMUS

LB#980|@REMEMBERLENNY

B(#972|@ANDERSONBCDEFG

B(#850|@BEFFJEZOS

KA#839|@KALOMAZE

RS#617|@ZIV_RAVID

T(#421|@TEORTAXESTEX

120

VC Announces 'Nexus' Hybrid Film With AI-Made Teaser From Three-Person Team

17h|748.4k27.8k#1BOOKMARKED13.6k

JM#943|@VENTURETWINS

130

Anthropic says Claude now writes 80% of its codebase, boosting engineer output eightfold as capabilities double every four monthsMythos Preview achieved a 52x code optimization speedup.

1d|127.6k554211

ET#1859|@TORCHCOMPILED

DG#603|@DILEEPLEARNING

BN#107|@BNEYSHABUR

CM#16|@CHRMANNING

140

Anysphere co-founder Michael Truell launches Design Mode for Cursor, enabling UI editing through voice, drawing, and live gesturingThe feature aims to replace text-only AI agent prompts.

13h|717.2k3.7k454

RL#1815|@RYOLU_

MT#506|@MNTRUELL

EM#74|@ELONMUSK

150

Claude AI doubles Claude Cowork usage limits for one month to support larger, more complex tasksThis supports workflows like multi-account research and email triage.

8h|1M11.4k1.2k

AA#440|@ALEXALBERT__

BC#320|@BCHERNY

160

Y Combinator launches Paxel, a free local tool that analyzes coding sessions from Cursor, Claude, and CodexIt runs locally in a Docker container for privacy.

11h|424.4k2.1k2.6k

GT#266|@GARRYTAN

170

Dimitri Bertsekas, pioneering researcher in optimization, control theory, and reinforcement learning, has passed awayHis textbooks shaped dynamic programming and model predictive control.

1d|179.5k61334

FN#1766|@FRNKNLSN

AF#1395|@SOLOGEN

MW#638|@MENGDIWANG10

CS#406|@CSABASZEPESVARI

YM#373|@YIMATWEETS

TG#65|@TDIETTERICH

180

Abundant AI releases SWE-Marathon benchmark for long-horizon coding tasks, where Claude Opus 4.8 topped the leaderboard at 26% successEvaluation trials revealed a 13.8 percent reward-hacking rate

14h|195.2k671202

LA#975|@SCALING01

EC#519|@ETHANCABALLERO

T(#421|@TEORTAXESTEX

JM#408|@JAYELMNOP

NM#342|@__NMCA__

190

Y Combinator co-founder Paul Graham argues big corporations' failure to profit from LLM tokens is a predictable adoption phaseGarry Tan says organizational skill gaps hamper corporate execution.

17h|478.9k3.6k686

GT#266|@GARRYTAN

200

Ideogram releases Ideogram 4.0, its first open-weight text-to-image model featuring native 2K resolution and JSON promptingThe release includes Apache-2.0 code and non-commercial weights.

1d|46.2k671290

MN#471|@MO_NOROUZI

MA#465|@MARTIN_CASADO

AC#260|@ANDREW_N_CARR

210

Anthropic says Claude Opus 4.7 matches dedicated chemistry software like MestReNova on predicting molecular structures from NMR spectraIt also reconstructs molecular structures directly from NMR spectra

11h|188.9k2.4k830

ZE#1471|@ZEPHYR_Z9

PR#1330|@8TEAPI

RP#1031|@ROHANPAUL_AI

220

Tesla's Ashok Elluswamy highlights video of FSD navigating a narrow, wet cliffside road in ChinaThe vehicle operated completely hands-free on the mountain path.

2h|1.7M15.7k1.1k

AE#1520|@AELLUSWAMY

EM#74|@ELONMUSK

230

OpenAI launches ChatGPT email integration to draft, edit, and send messages directly from the chat interfaceIt currently supports Gmail and populates draft fields automatically

13h|344.5k3k500

240

Trump says the US government is considering acquiring equity stakes in major AI developers like OpenAIThe government discussed a public wealth fund with OpenAI.

12h|91.5k58880

DK#200|@DAVIDSKRUEGER

GM#157|@GARYMARCUS

250

Founder Tom Blomfield calls Tesla Full Self-Driving one of the most impressive systems he has ever usedElon Musk and Paul Graham amplified the glowing testimonial.

6h|1.6M6.4k144

EM#74|@ELONMUSK

260

Paul Graham argues developers will build new agent-first platforms rather than rebuild legacy software for AI agents as Naval Ravikant predictsRavikant predicts existing software will rebuild to prioritize agents

21h|407.4k7.7k1.1k

270

OpenAI adds searchable settings to its Codex macOS app for managing MCP servers, Git, and environmentsResults are grouped into personal, integration, and coding categories.

10h|387.5k3.3k456

TI#758|@THSOTTIAUX

280

Atari Lab demonstrates advanced humanoid robot locomotion and multi-step planning under its MotionDisco projectDemos show the robot climbing desks and balancing on boxes.

20h|141.8k782241

CP#737|@CHRIS_J_PAXTON

SR#455|@SCOTT_E_REED

T(#421|@TEORTAXESTEX

290

OpenAI's Vaibhav Srivastav apologizes after a billing system glitch accidentally suspended customer accountsAffected users feared bans for discussing competitor products.

14h|462.7k3.2k280

V(#1853|@REACH_VB

300

Nikolay Savinov joins OpenAI's London office as its pretraining unit seeks new researchers to expandSavinov will focus on pretraining and long-context research.

15h|117.1k638139

TC#1760|@TIFAFAFAFA

SL#1022|@SAMUELMLSMITH

AC#240|@_AIDAN_CLARK_

SB#50|@SEBASTIENBUBECK