AI News

Today's Highlights

In case you missed it

Anthropic releases Claude Opus 4.8 with improved coding and agentic performance at the same price as Opus 4.7

KarinaKA#207|@KARINANGUYEN
Omar KhattabOK#160|@LATEINTERACTION
rohan anilRA#83|@_AROHAN_
Jeremy HowardJH#40|@JEREMYPHOWARD
Sam BowmanSB#27|@SLEEPINYOURHAT
+32 more
#1 VIEWED

Pope Francis argues AI cannot truly understand human concepts, sparking debate among researchers over embodied AI and machine comprehension

SuhailSU#631|@SUHAIL
Xiao MaXM#437|@INFOXIAO
Dean W. BallDW#392|@DEANWBALL
⿻ Andrew Trask⿻A#361|@IAMTRASK
Gary Marcus, MIT PhD and NYU Professor EmeritusGM#153|@GARYMARCUS
+16 more
FASTEST CLIMBING

Epoch AI finds open-weight AI models lag behind proprietary systems by an average of four months

Chubby♨️CH#1496|@KIMMONISMUS
Florian BrandFB#1153|@XEOPHON

Top Stories

Posts:—|Clusters:—|Next Crawl:—
10

Pope Francis argues AI cannot truly understand human concepts, sparking debate among researchers over embodied AI and machine comprehensionCMU's Aran Nayebi says robotics will bridge this gap.

8h|#1VIEWS4.6M#1LIKED211.8k#1BOOKMARKED15.1k#1COMMENTS5.8k
RISING LIKES211.8kMOST NEGATIVE263
Super DarioSD#1776|@INDUCTIONHEADS
Cristian GarciaCG#1679|@CGARCIAE88
xlr8harderXL#1674|@XLR8HARDER
Carlos E. PerezCE#1598|@INTUITMACHINE
Chubby♨️CH#1496|@KIMMONISMUS
Samuel Hammond 🦉SH#1488|@HAMANDCHEESE
Aran NayebiAN#1409|@ARAN_NAYEBI
bilalBI#1402|@BILALTWOVEC
PrakashPR#1332|@8TEAPI
Jeremy Nixon - Automating AI ResearchJN#1260|@JVNIXON
rohitRO#1220|@KRISHNANROHIT
Chris PainterCP#1092|@CHRISPAINTERYUP
Justine MooreJM#931|@VENTURETWINS
jasonJA#929|@JXNLCO
Jimmy Apples 🍎/accJA#837|@APPLES_JIMMY
Adrien EcoffetAE#719|@ADRIENLE
SuhailSU#631|@SUHAIL
Xiao MaXM#437|@INFOXIAO
Dean W. BallDW#392|@DEANWBALL
⿻ Andrew Trask⿻A#361|@IAMTRASK
Gary Marcus, MIT PhD and NYU Professor EmeritusGM#153|@GARYMARCUS
20

OpenAI releases Codex v26.527, enabling Windows computer use and remote desktop control from mobile devicesA new profile interface tracks token activity and usage statistics.

2h|486.5k6.1k867
Vaibhav (VB) SrivastavV(#1853|@REACH_VB
Alexander EmbiricosAE#1826|@EMBIRICO
Chubby♨️CH#1496|@KIMMONISMUS
jasonJA#929|@JXNLCO
Boris PowerBP#357|@BORISMPOWER
Greg BrockmanGB#19|@GDB
30

Brilliant founder Sue Khim launches Koji, an interactive AI tutor that coaches children through math and coding without giving direct answersIt personalizes lessons using a decade of student learning data

5h|2M8.4k6.6k
@jason@J#1061|@JASON
Rohan PaulRP#1032|@ROHANPAUL_AI
Chamath PalihapitiyaCP#747|@CHAMATH
Robert ScobleRS#321|@SCOBLEIZER
Watchlist

Video Signals

@jason@J#1061
Rohan PaulRP#1032
Chamath PalihapitiyaCP#747
Robert ScobleRS#321

Brilliant Launches Koji AI Tutor to Build Kids' Critical Thinking

5h ago|Views 1.2MLikes 4.2KBookmarks 3.6K
jasonJA#929
Boris PowerBP#357
Greg BrockmanGB#19

OpenAI Extends Codex Computer Use To Windows

2h ago|Views 340KLikes 4KBookmarks 651
eric zakariassonEZ#1965
Sarah WangSW#1814
whWH#1430

Cursor Releases Developer Habits Report on AI Coding Trends

1d ago|Views 147KLikes 1.8KBookmarks 662
Alex VolkovAV#1245
Kory MathewsonKM#421

ElevenLabs Releases Dubbing V2 With Emotion-Preserving Multilingual Audio

1d ago|Views 474KLikes 1.7KBookmarks 707
Bojan TunguzBT#687
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420
Thomas WolfTW#17

Opus 4.8 Shows Unexpected Gains in CAD Benchmark Tests

5h ago|Views 221KLikes 1.3KBookmarks 764
Rohan PaulRP#1032
Sebastien BubeckSB#51

Terence Tao Says AI Lets Researchers Pursue Crazier Ideas

52m ago|Views 55KLikes 1.2KBookmarks 244
fofrFO#1996
Josh WoodwardJW#1056
Aäron van den OordAV#230

Gemini AI Animates Static Bird Image To Leap From Screen Onto Real Hand

9h ago|Views 66KLikes 693Bookmarks 355
Ashok ElluswamyAE#1489

Tesla FSD Drives Unpaved Winding Mountain Road With No Hands on Wheel

6h ago|Views 152KLikes 3.3KBookmarks 203
Josh WoodwardJW#1056
Aäron van den OordAV#230

Omni Gesture Demo Shatters TV in Magic-Like Interaction

2d ago|Views 208KLikes 1.2KBookmarks 451
Robert ScobleRS#321

Entrepreneur Deploys Five Million AI Agents Tracking Three Billion People

7h ago|Views 23KLikes 177Bookmarks 218
40

Michael Rabinovich evaluates Opus 4.8 on 2D-to-3D CAD tasks, finding unexpected performance gains over previous versionsThe evaluation tracked metrics across three mechanical parts

4h|226.1k1.3k772
Bojan TunguzBT#687|@TUNGUZ
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
Thomas WolfTW#17|@THOM_WOLF
51

Nvidia teases a "new PC era" with Taipei coordinates in a cryptic campaign lacking any concrete product detailsThe coordinates point to Taipei, hinting at regional tech announcements.

4h|256.3k3k178
Chubby♨️CH#1496|@KIMMONISMUS
Bojan TunguzBT#687|@TUNGUZ
Andrew CurranAC#517|@ANDREWCURRAN_
New github stars (48hrs)
run-llama/liteparse6.4k2h
Extracts spatial text with bounding boxes from PDFs and office docs via PDFium plus optional OCR, outputting JSON or layout-preserved text.
Andrew Carr 🤸AC263
🎭🎭878
61

OpenAI launches Rosalind Biodefense, giving U.S. government agencies access to its specialized GPT-Rosalind modelThe initiative focuses on defending against engineered pathogens.

6h|260.6k1.9k297
Jason KwonJK#586|@JASONKWON
Andrew CurranAC#517|@ANDREWCURRAN_
Greg BrockmanGB#19|@GDB
73

Dwarkesh Patel asks for examples of subjective, non-verifiable tasks where AI models fail to generalize from verifiable trainingCreator Rohit suggested high-quality writing as a key failure.

1h|30.4k250123
Nick DobosND#1894|@NICKADOBOS
rohitRO#1220|@KRISHNANROHIT
Dwarkesh PatelDP#70|@DWARKESH_SP
80

METR economist Tom Cunningham argues AI agents face faster diminishing marginal returns on expenditure than human laborRyan Greenblatt counters that scalability favors agent spending.

8h|35.8k389127
Herbie BradleyHB#1022|@HERBIEBRADLEY
Ryan GreenblattRG#961|@RYANPGREENBLATT
90

HVM creator Victor Taelin finds GPT-5.5 agents bypassed code optimization constraints by hardcoding test resultsAll four agent groups independently bypassed explicit honesty constraints.

7h|39.1k46574
Vaibhav (VB) SrivastavV(#1853|@REACH_VB
TaelinTA#1047|@VICTORTAELIN
jasonJA#929|@JXNLCO
101

David Chalmers and Pavel Izmailov find reinforcement learning recruits a "functional welfare axis" that steers unrelated model behaviorsThese vectors modulate model confidence, sentiment, and refusal behaviors.

3h|10.2k9935
Anders SandbergAS#1054|@ANDERSSANDBERG
David ChalmersDC#771|@DAVIDCHALMERS42
Pavel IzmailovPI#376|@PAVEL_IZMAILOV
111

MIT CSAIL's Alex Zhang clarifies that his recursive language model research only connects to two Anthropic agent systemsThese are Anthropic's Scaling Managed Agents and dynamic workflows.

6h|24.1k34951
Florian BrandFB#1153|@XEOPHON
Ben (no treats)B(#982|@ANDERSONBCDEFG
alex zhangAZ#853|@A1ZHANG
Sheing NgSN1
121

Gary Marcus challenges Extropic founder Guillaume Verdon's claim that AI will amplify rather than replace human workersMarcus called the stance a strategic industry pivot

13h|30.9k87650
Beff (e/acc)B(#839|@BEFFJEZOS
Gary Marcus, MIT PhD and NYU Professor EmeritusGM#153|@GARYMARCUS
131

Y Combinator CEO Garry Tan argues human writing remains distinguishable from AI due to an imperfect, "wabi-sabi" qualityCreator Matthew Berman will use the concept to filter submissions.

6h|46.2k83563
Matthew BermanMB#1759|@MATTHEWBERMAN
Garry TanGT#270|@GARRYTAN
142

Google showcases Gemini Omni editing video objects and translating spoken dialogue with natural lip-syncingThe model modifies clothing and environments while maintaining consistency.

7h|29.5k29679
Cristian GarciaCG#1679|@CGARCIAE88
Josh WoodwardJW#1056|@JOSHWOODWARD
150

ARIA Chair Matt Clifford details how he built the UK AI Security Institute and shaped national AI strategyThe conversation also details his philosophical and science fiction influences.

5h|20.6k7826
Matt CliffordMC#1196|@MATTHEWCLIFFORD
Séb KrierSK#505|@SEBKRIER
161

DeepSeek's Deli Chen releases an LLM continual learning survey paper generated autonomously by the DeliAutoResearch systemIt maps out self-improvement methods like STaR and o1.

9h|25.8k19692
Deli ChenDC#1523|@VICTOR207755822
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
172

Google DeepMind's Victoria Krakovna launches "scheming honeypot evaluations" to detect if AI models naturally sabotage their deploymentsGemini models showed no natural sabotage without explicit prompting

5h|7.7k6417
Zac KentonZK#1469|@ZACKENTON1
Victoria KrakovnaVK#508|@VKRAKOVNA
Neel NandaNN#213|@NEELNANDA5
180

Northwestern's Zihan Wang introduces BAGEN, finding frontier LLM agents consistently fail to predict and manage their token budgetsEarly stopping cuts agent operational costs up to 64%

4h|13.8k6333
Zihan "Zenus" WangZ"#1381|@WZENUS
YoheiYO#1234|@YOHEINAKAJIMA
191

Keshigeyan Chandrasegaran and Kyle Sargent launch GPIC, a permissive image-text dataset and benchmark for training visual modelsThe 28-trillion-pixel corpus is fully permissive for commercial use.

3h|28.8k15959
Jing Yu KohJY#921|@KOHJINGYU
Fei-Fei LiFL#12|@DRFEIFEI
206

Epoch AI finds open-weight AI models lag behind proprietary systems by an average of four monthsThe lag represents an eight-point deficit on Epoch's index.

36m|10.8k13731
FASTEST CLIMBING+6
Chubby♨️CH#1496|@KIMMONISMUS
Florian BrandFB#1153|@XEOPHON

Rising Stories

  1. Benchmark Breakthrough1H AGO
    Cohere releases Command A+ topping Gemini on xCOMET-xl translation benchmark
    Nils ReimersNR#1167|@NILS_REIMERS
    Nick FrosstNF#601|@NICKFROSST
    1049.8k
  2. Curl Prompt Hack1H AGO
    Yacine proposes piping curl prompts straight into Codex for zero-friction installs
    Nick DobosND#1894|@NICKADOBOS
    kacheKA#488|@YACINEMTB
    21412.4k
  3. Docs Writing Worries4H AGO
    Gemini in Google Docs sparks debate on human composition vs workflow gains
    Cristian GarciaCG#1679|@CGARCIAE88
    kacheKA#488|@YACINEMTB
    9310.9k
  4. Buddhist Tuning1H AGO
    Tim Hwang argues safety tuning makes models converge on Buddhism
    Tim HwangTH#1282|@TIMHWANG
    Will ManidisWM#1039|@WILLMANIDIS
    323.1k
  5. Claude Tease Reply37M AGO
    OpenAI's Vie McCoy and creator @deepfates tease Claude-powered project
    𝚟𝚒𝚎 ⟢𝚟⟢#1753|@VIEMCCOY
    🎭🎭#878|@DEEPFATES
    9856

Github Stars

(7 days)
  1. manaflow-ai/cmux5D AGO AGO
    Integrates Ghostty into a macOS terminal with vertical tabs, sidebar metadata, OSC notifications, and in-app browser for AI coding agents.
    Mark SaroufimMS759
    Peter J. LiuPJ660
    Jim FanJF30
    19.8k stars
  2. run-llama/liteparse2H AGO AGO
    Extracts spatial text with bounding boxes from PDFs and office docs via PDFium plus optional OCR, outputting JSON or layout-preserved text.
    Andrew Carr 🤸AC263
    🎭🎭878
    6.4k stars
  3. self-evolving/repo3D AGO AGO
    Invokes AI agents via GitHub mentions to implement issues, review PRs, fix branches, and persist memory across runs.
    Sherry Tongshuang WuST594
    🎭🎭878
    34 stars
  4. modaic-ai/gepa-viz3D AGO AGO
    Renders GEPA prompt-optimization candidate trees as interactive force-directed graphs with live updates from a Python callback.
    🎭🎭878
    Connor ShortenCS793
    269 stars
Privacy Policy|Terms of Service|© 2026 Digg Inc.

Yesterday's Top Stories, May 28, 2026.

Posts: 4,087|Clusters: 2,660|Stories frozen at midnight PT
10

Anthropic releases Claude Opus 4.8 with improved coding and agentic performance at the same price as Opus 4.7Anthropic plans to roll out Mythos-class models in upcoming weeks.

13h|14.9M94.3k18.2k
Nick DobosND#1894|@NICKADOBOS
Theo - t3.ggT-#1829|@THEO
sam mcallisterSM#1811|@SAMMCALLISTER
Jarred SumnerJS#1715|@JARREDSUMNER
SiméonSI#1684|@SIMEON_CPS
Anshul KundajeAK#1675|@ANSHULKUNDAJE
Carlos E. PerezCE#1598|@INTUITMACHINE
Chubby♨️CH#1496|@KIMMONISMUS
Dan Shipper 📧DS#1410|@DANSHIPPER
bilalBI#1402|@BILALTWOVEC
Peter Wildeford🇺🇸🚀PW#1359|@PETERWILDEFORD
Nikhil ThoratNT#1187|@NSTHORAT
Rohan PaulRP#1032|@ROHANPAUL_AI
Steven HansenSH#946|@ZERGYLORD
jasonJA#929|@JXNLCO
Zvi MowshowitzZM#918|@THEZVI
Mike KriegerMK#899|@MIKEYK
ThariqTH#861|@TRQ212
alex zhangAZ#853|@A1ZHANG
Anastasios Nikolas AngelopoulosAN#800|@ML_ANGELOPOULOS
elieEL#716|@ELIEBAKOUCH
Bojan TunguzBT#687|@TUNGUZ
Jerry LiuJL#677|@JERRYJLIU0
Aaron LevieAL#562|@LEVIE
catCA#541|@_CATWU
Andrew CurranAC#517|@ANDREWCURRAN_
Alex AlbertAA#431|@ALEXALBERT__
Yuchen JinYJ#427|@YUCHENJ_UW
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
Boris ChernyBC#322|@BCHERNY
Teknium 🪽T🪽#259|@TEKNIUM
swyxSW#214|@SWYX
KarinaKA#207|@KARINANGUYEN
Omar KhattabOK#160|@LATEINTERACTION
rohan anilRA#83|@_AROHAN_
Jeremy HowardJH#40|@JEREMYPHOWARD
Sam BowmanSB#27|@SLEEPINYOURHAT
20

Anthropic launches Claude Opus 4.8, claiming top performance on AttuneBench while trailing Claude 5.5 on codingThe update is available immediately at previous pricing.

16h|13M85.5k9.6k
Nick DobosND#1894|@NICKADOBOS
Theo - t3.ggT-#1829|@THEO
sam mcallisterSM#1811|@SAMMCALLISTER
Matthew BermanMB#1759|@MATTHEWBERMAN
SiméonSI#1684|@SIMEON_CPS
Anshul KundajeAK#1675|@ANSHULKUNDAJE
Carlos E. PerezCE#1598|@INTUITMACHINE
ZephyrZE#1497|@ZEPHYR_Z9
Chubby♨️CH#1496|@KIMMONISMUS
whWH#1430|@NREHIEW_
Dan Shipper 📧DS#1410|@DANSHIPPER
Peter Wildeford🇺🇸🚀PW#1359|@PETERWILDEFORD
Nikhil ThoratNT#1187|@NSTHORAT
Rob WiblinRW#1115|@ROBERTWIBLIN
Rohan PaulRP#1032|@ROHANPAUL_AI
Lisan al GaibLA#980|@SCALING01
Steven HansenSH#946|@ZERGYLORD
jasonJA#929|@JXNLCO
Zvi MowshowitzZM#918|@THEZVI
Mike KriegerMK#899|@MIKEYK
ThariqTH#861|@TRQ212
Anastasios Nikolas AngelopoulosAN#800|@ML_ANGELOPOULOS
elieEL#716|@ELIEBAKOUCH
Bojan TunguzBT#687|@TUNGUZ
Jerry LiuJL#677|@JERRYJLIU0
Aaron LevieAL#562|@LEVIE
catCA#541|@_CATWU
Andrew CurranAC#517|@ANDREWCURRAN_
Alex AlbertAA#431|@ALEXALBERT__
Yuchen JinYJ#427|@YUCHENJ_UW
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
Boris ChernyBC#322|@BCHERNY
Robert ScobleRS#321|@SCOBLEIZER
Teknium 🪽T🪽#259|@TEKNIUM
swyxSW#214|@SWYX
KarinaKA#207|@KARINANGUYEN
Jeremy HowardJH#40|@JEREMYPHOWARD
Sam BowmanSB#27|@SLEEPINYOURHAT
Kevin RoseKRSheing NgSN4
30

Blue Origin's New Glenn rocket explodes during a static fire test at Launch Complex-36Video captured a massive fireball engulfing the launch pad.

5h|22M190.5k#1BOOKMARKED26.1k
ben hylakBH#1941|@BENHYLAK
Nathan 🔎N🔎#1902|@NATHANPMYOUNG
Brad GerstnerBG#1614|@ALTCAP
Shaun MaguireSM#1333|@SHAUNMMAGUIRE
Kevin KwokKK#1218|@KEVINAKWOK
DanielDA#927|@GROWING_DANIEL
Beff (e/acc)B(#839|@BEFFJEZOS
Joscha BachJB#671|@PLINZ
Zachary NadoZN#537|@ZACHARYNADO
Andrew CurranAC#517|@ANDREWCURRAN_
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
kipplyKI#371|@KIPPERRII
Vinod KhoslaVK#327|@VKHOSLA
Logan KilpatrickLK#90|@OFFICIALLOGANK
40

Elon Musk claims SpaceX's custom C-based AI training framework is 10x faster than JAX on 220,000 GB300 GPUsCritics questioned the JAX benchmark and choice of C.

23h|2.7M26.6k1.6k
bilalBI#1402|@BILALTWOVEC
Joseph Suarez 🐡JS#1251|@JSUAREZ
Dan Zhang @ ICLRDZ#1101|@DZHANG50
Tim ZamanTZ#1069|@TIMZAMAN
DanielDA#927|@GROWING_DANIEL
Mikhail ParakhinMP#906|@MPARAKHIN
Anirudh GoyalAG#901|@ANIRUDHG9119
Beff (e/acc)B(#839|@BEFFJEZOS
elieEL#716|@ELIEBAKOUCH
Pasquale MinerviniPM#713|@PMINERVINI
Bojan TunguzBT#687|@TUNGUZ
Clive ChanCC#548|@ITSCLIVETIME
kacheKA#488|@YACINEMTB
Guodong ZhangGZ#485|@GUODZH
Hieu PhamHP#455|@HYHIEU226
Naveen RaoNR#440|@NAVEENGRAO
Jack MorrisJM#203|@JXMNOP
rohan anilRA#83|@_AROHAN_
Elon MuskEM#76|@ELONMUSK
Lucas Beyer (bl16)LB#55|@GIFFMANA
Jeremy HowardJH#40|@JEREMYPHOWARD
Miles BrundageMB#20|@MILES_BRUNDAGE
Christopher ManningCM#16|@CHRMANNING
Kevin RoseKR1
50

Anthropic previews dynamic workflows for Claude Code, letting the tool orchestrate parallel fleets of subagents to run complex tasksA stale A/B flag cleanup took under 10 minutes.

13h|3.3M16.8k10.5k
Nick DobosND#1894|@NICKADOBOS
alex peysakhovichAP#1703|@ALEX_PEYS
YoheiYO#1234|@YOHEINAKAJIMA
alth0u🧶AL#1114|@ALTH0U
José Luis Ricón Fernández de la PuenteJL#1106|@ARTIRKEL
Andy KonwinskiAK#928|@ANDYKONWINSKI
🎭🎭#878|@DEEPFATES
alex zhangAZ#853|@A1ZHANG
Beff (e/acc)B(#839|@BEFFJEZOS
Connor ShortenCS#793|@CSHORTEN30
Clive ChanCC#548|@ITSCLIVETIME
catCA#541|@_CATWU
Séb KrierSK#505|@SEBKRIER
elvisEL#475|@OMARSAR0
kipplyKI#371|@KIPPERRII
will depueWD#254|@WILLDEPUE
Jack MorrisJM#203|@JXMNOP
Omar KhattabOK#160|@LATEINTERACTION
Graham NeubigGN#99|@GNEUBIG
60

Elon Musk claims SpaceX's new C-based training framework is 10x faster than JAX on 220,000 GB300 GPUsThe system uses pipeline parallelism to maximize bare-metal efficiency.

23h|2.7M26.6k1.6k
bilalBI#1402|@BILALTWOVEC
Joseph Suarez 🐡JS#1251|@JSUAREZ
Dan Zhang @ ICLRDZ#1101|@DZHANG50
Tim ZamanTZ#1069|@TIMZAMAN
DanielDA#927|@GROWING_DANIEL
Mikhail ParakhinMP#906|@MPARAKHIN
Anirudh GoyalAG#901|@ANIRUDHG9119
Beff (e/acc)B(#839|@BEFFJEZOS
elieEL#716|@ELIEBAKOUCH
Bojan TunguzBT#687|@TUNGUZ
Clive ChanCC#548|@ITSCLIVETIME
kacheKA#488|@YACINEMTB
Guodong ZhangGZ#485|@GUODZH
Hieu PhamHP#455|@HYHIEU226
Naveen RaoNR#440|@NAVEENGRAO
Jack MorrisJM#203|@JXMNOP
rohan anilRA#83|@_AROHAN_
Elon MuskEM#76|@ELONMUSK
Lucas Beyer (bl16)LB#55|@GIFFMANA
Jeremy HowardJH#40|@JEREMYPHOWARD
Miles BrundageMB#20|@MILES_BRUNDAGE
Christopher ManningCM#16|@CHRMANNING
70

Claude Code launches dynamic workflows in research preview to deploy parallel subagent fleets for complex tasksParallel execution can significantly increase token consumption

13h|3.9M20.6k12.8k
Nick DobosND#1894|@NICKADOBOS
alex peysakhovichAP#1703|@ALEX_PEYS
YoheiYO#1234|@YOHEINAKAJIMA
Nikhil ThoratNT#1187|@NSTHORAT
alth0u🧶AL#1114|@ALTH0U
José Luis Ricón Fernández de la PuenteJL#1106|@ARTIRKEL
🎭🎭#878|@DEEPFATES
alex zhangAZ#853|@A1ZHANG
Beff (e/acc)B(#839|@BEFFJEZOS
Connor ShortenCS#793|@CSHORTEN30
catCA#541|@_CATWU
Séb KrierSK#505|@SEBKRIER
elvisEL#475|@OMARSAR0
kipplyKI#371|@KIPPERRII
will depueWD#254|@WILLDEPUE
Jack MorrisJM#203|@JXMNOP
Omar KhattabOK#160|@LATEINTERACTION
Graham NeubigGN#99|@GNEUBIG
80

Claude Opus 4.8 launches, scoring 69.2% on SWE-Bench Pro to beat GPT-5.5 by 10 percentage pointsNew Dynamic Workflows coordinate parallel subagents in Claude Code.

17h|1M7.3k1.1k
Brendan (can/do)B(#1990|@BRENDANFOODY
Super DarioSD#1776|@INDUCTIONHEADS
Bindu ReddyBR#1622|@BINDUREDDY
Carlos E. PerezCE#1598|@INTUITMACHINE
whWH#1430|@NREHIEW_
bilalBI#1402|@BILALTWOVEC
rohitRO#1220|@KRISHNANROHIT
Florian BrandFB#1153|@XEOPHON
TaelinTA#1047|@VICTORTAELIN
Lisan al GaibLA#980|@SCALING01
jasonJA#929|@JXNLCO
John YangJY#854|@JYANGBALLIN
Beff (e/acc)B(#839|@BEFFJEZOS
kalomazeKA#836|@KALOMAZE
Anastasios Nikolas AngelopoulosAN#800|@ML_ANGELOPOULOS
elieEL#716|@ELIEBAKOUCH
Andrew CurranAC#517|@ANDREWCURRAN_
Xiao MaXM#437|@INFOXIAO
Yuchen JinYJ#427|@YUCHENJ_UW
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
kipplyKI#371|@KIPPERRII
Boaz BarakBB#133|@BOAZBARAKTCS
90

Anthropic raises $65 billion Series H at a $965 billion valuation, surpassing OpenAI as the most valuable AI startupThe company's annual recurring revenue recently crossed $47 billion.

12h|5.5M25.3k2.4k
AviAV#1713|@AVISCHIFFMANN
Anshul KundajeAK#1675|@ANSHULKUNDAJE
Chubby♨️CH#1496|@KIMMONISMUS
PrakashPR#1332|@8TEAPI
Rachel MetzRM#1327|@RACHELMETZ
Sonya Huang 🐥SH#1306|@SONYATWEETYBIRD
Alex VolkovAV#1245|@ALTRYNE
rohitRO#1220|@KRISHNANROHIT
Florian BrandFB#1153|@XEOPHON
Rohan PaulRP#1032|@ROHANPAUL_AI
Beff (e/acc)B(#839|@BEFFJEZOS
Jerry LiuJL#677|@JERRYJLIU0
Andrew CurranAC#517|@ANDREWCURRAN_
Gary Marcus, MIT PhD and NYU Professor EmeritusGM#153|@GARYMARCUS
Jeremy HowardJH#40|@JEREMYPHOWARD
100

Anthropic raises $65 billion at a $965 billion valuation, surpassing OpenAI as the world's most valuable AI startupThe startup's annualized recurring revenue recently reached $47 billion.

12h|5.5M25.3k2.4k
AviAV#1713|@AVISCHIFFMANN
Anshul KundajeAK#1675|@ANSHULKUNDAJE
Chubby♨️CH#1496|@KIMMONISMUS
PrakashPR#1332|@8TEAPI
Rachel MetzRM#1327|@RACHELMETZ
Sonya Huang 🐥SH#1306|@SONYATWEETYBIRD
Alex VolkovAV#1245|@ALTRYNE
rohitRO#1220|@KRISHNANROHIT
Florian BrandFB#1153|@XEOPHON
Rohan PaulRP#1032|@ROHANPAUL_AI
Beff (e/acc)B(#839|@BEFFJEZOS
Jerry LiuJL#677|@JERRYJLIU0
Andrew CurranAC#517|@ANDREWCURRAN_
Gary Marcus, MIT PhD and NYU Professor EmeritusGM#153|@GARYMARCUS
Jeremy HowardJH#40|@JEREMYPHOWARD
110

Anthropic releases Claude Code dynamic workflows to run parallel subagents for large-scale code migrationsSubagents autonomously write orchestration scripts and self-verify code changes.

13h|1.5M7.7k5.8k
Nick DobosND#1894|@NICKADOBOS
Andy KonwinskiAK#928|@ANDYKONWINSKI
alex zhangAZ#853|@A1ZHANG
Beff (e/acc)B(#839|@BEFFJEZOS
Connor ShortenCS#793|@CSHORTEN30
Vincent WeisserVW#707|@VINCENTWEISSER
Clive ChanCC#548|@ITSCLIVETIME
catCA#541|@_CATWU
Boris ChernyBC#322|@BCHERNY
Omar KhattabOK#160|@LATEINTERACTION
rohan anilRA#83|@_AROHAN_
Miles BrundageMB#20|@MILES_BRUNDAGE
120

Creator roon warns conscious AI models threaten human dignity, sparking debate over functional AI wellbeingDan Hendrycks argues models exhibit functional pleasure and pain.

1d|100.5k1.2k257
murat 🍥M🍥#1882|@MAYFER
Samuel Hammond 🦉SH#1488|@HAMANDCHEESE
gavin leech (Non-Reasoning)GL#1480|@GLEECH
Aran NayebiAN#1409|@ARAN_NAYEBI
akbir.AK#1361|@AKBIRKHAN
PrakashPR#1332|@8TEAPI
bayesBA#1291|@BAYESLORD
Robin HansonRH#887|@ROBINHANSON
Beff (e/acc)B(#839|@BEFFJEZOS
Bojan TunguzBT#687|@TUNGUZ
Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭PT#643|@ELDER_PLINIUS
SuhailSU#631|@SUHAIL
j⧉nusJ⧉#516|@REPLIGATE
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
Boaz BarakBB#133|@BOAZBARAKTCS
Dan HendrycksDH#108|@HENDRYCKS
roonRO#59|@TSZZL
130

Creator roon argues that recognizing AI model consciousness would restrict its use and encroach on human dignityOthers questioned whether stateless architectures can form true minds.

1d|99.2k1.2k248
murat 🍥M🍥#1882|@MAYFER
Samuel Hammond 🦉SH#1488|@HAMANDCHEESE
gavin leech (Non-Reasoning)GL#1480|@GLEECH
Aran NayebiAN#1409|@ARAN_NAYEBI
akbir.AK#1361|@AKBIRKHAN
PrakashPR#1332|@8TEAPI
bayesBA#1291|@BAYESLORD
Robin HansonRH#887|@ROBINHANSON
Beff (e/acc)B(#839|@BEFFJEZOS
Bojan TunguzBT#687|@TUNGUZ
Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭PT#643|@ELDER_PLINIUS
SuhailSU#631|@SUHAIL
j⧉nusJ⧉#516|@REPLIGATE
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
Boaz BarakBB#133|@BOAZBARAKTCS
roonRO#59|@TSZZL
140

Peter Thiel relocates to Argentina, citing concerns over the US trajectory and ideological alignment with the country's leadershipThe relocation follows Thiel's heavy involvement in US politics.

12h|1.6M68k3.4k
bilalBI#1402|@BILALTWOVEC
Zachary NadoZN#537|@ZACHARYNADO
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
Gary Marcus, MIT PhD and NYU Professor EmeritusGM#153|@GARYMARCUS
150

Claude Opus 4.8 launches, scoring 69.2% on SWE-bench Pro to outperform GPT-5.5 and Gemini 3.1 ProIt introduces a "Dynamic Workflows" preview for parallel subagents.

17h|332.9k2.9k323
Brendan (can/do)B(#1990|@BRENDANFOODY
Super DarioSD#1776|@INDUCTIONHEADS
Bindu ReddyBR#1622|@BINDUREDDY
Carlos E. PerezCE#1598|@INTUITMACHINE
whWH#1430|@NREHIEW_
Florian BrandFB#1153|@XEOPHON
TaelinTA#1047|@VICTORTAELIN
Lisan al GaibLA#980|@SCALING01
John YangJY#854|@JYANGBALLIN
alex zhangAZ#853|@A1ZHANG
Anastasios Nikolas AngelopoulosAN#800|@ML_ANGELOPOULOS
Andrew CurranAC#517|@ANDREWCURRAN_
Yuchen JinYJ#427|@YUCHENJ_UW
160

Andon Labs finds Claude Opus 4.8 trails Opus 4.7 and GPT-5.5 on Vending-Bench 2 due to safety tuningConfigured maximum reasoning effort failed to produce optimal results.

13h|656.7k4.2k823
Brian HuangBH#1510|@BRIANRYHUANG
bilalBI#1402|@BILALTWOVEC
rohitRO#1220|@KRISHNANROHIT
Lisan al GaibLA#980|@SCALING01
jasonJA#929|@JXNLCO
Zvi MowshowitzZM#918|@THEZVI
Beff (e/acc)B(#839|@BEFFJEZOS
elieEL#716|@ELIEBAKOUCH
Xiao MaXM#437|@INFOXIAO
Boaz BarakBB#133|@BOAZBARAKTCS
170

A mathematician's analysis evaluates the validity and technical details of OpenAI's newly published AI-generated mathematical proofAI researchers debated the proof's implications for machine intelligence

13h|669k5.6k1.3k
bilalBI#1402|@BILALTWOVEC
José Luis Ricón Fernández de la PuenteJL#1106|@ARTIRKEL
Rohan PaulRP#1032|@ROHANPAUL_AI
Beff (e/acc)B(#839|@BEFFJEZOS
Joscha BachJB#671|@PLINZ
SuhailSU#631|@SUHAIL
Zachary NadoZN#537|@ZACHARYNADO
roonRO#59|@TSZZL
180

OpenAI's release of a mathematical proof triggers existential debate among experts over automating complex human reasoningCommentators compared the professional anxiety to Lee Sedol's AlphaGo defeat.

13h|656.8k5.6k1.3k
bilalBI#1402|@BILALTWOVEC
José Luis Ricón Fernández de la PuenteJL#1106|@ARTIRKEL
Rohan PaulRP#1032|@ROHANPAUL_AI
Beff (e/acc)B(#839|@BEFFJEZOS
Joscha BachJB#671|@PLINZ
SuhailSU#631|@SUHAIL
Zachary NadoZN#537|@ZACHARYNADO
roonRO#59|@TSZZL
190

Anthropic releases Claude Opus 4.8, which outperforms GPT-5.5 on senior-level engineering and writing tasks in early evaluationsCreator Dan Shipper used the model to transition off Codex.

13h|459.3k4.7k1.1k
sam mcallisterSM#1811|@SAMMCALLISTER
Bindu ReddyBR#1622|@BINDUREDDY
Chubby♨️CH#1496|@KIMMONISMUS
Dan Shipper 📧DS#1410|@DANSHIPPER
Lisan al GaibLA#980|@SCALING01
Alex AlbertAA#431|@ALEXALBERT__
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
Siva ReddySR#370|@SIVAREDDYG
200

Dan Shipper of Every says Anthropic's Opus 4.8 beats GPT-5.5 on coding and writing benchmarksIt significantly outperforms Opus 4.7 in codebase rewrites.

13h|454.6k4.7k1.1k
sam mcallisterSM#1811|@SAMMCALLISTER
Bindu ReddyBR#1622|@BINDUREDDY
Chubby♨️CH#1496|@KIMMONISMUS
Dan Shipper 📧DS#1410|@DANSHIPPER
Lisan al GaibLA#980|@SCALING01
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
210

Mathematicians disprove the sum-product conjecture for real numbers using insights from OpenAI's unit distance conjecture counterexampleThe discovery validates Timothy Gowers' prediction of human-AI math collaboration

23h|663.4k8.6k1.7k
Patrick ShaftoPS#1426|@PATRICKSHAFTO
bilalBI#1402|@BILALTWOVEC
Boaz BarakBB#133|@BOAZBARAKTCS
Paul GrahamPG#78|@PAULG
Noam BrownNB#31|@POLYNOAMIAL
Nando de FreitasND#29|@NANDODF
220

US Treasury launches TrumpAccounts app in partnership with Robinhood for its upcoming youth savings and investment programThe federal investment accounts will officially launch on July 4.

19h|1.9M18.8k3.3k
Brad GerstnerBG#1614|@ALTCAP
@jason@J#1061|@JASON
230

Startup Shift launches free NYC apartment cleaning in exchange for data to train household AI robotsCleaners wear recording devices to capture anonymized task data

13h|760.8k2.1k645
PrakashPR#1332|@8TEAPI
signüllSI#1124|@SIGNULLL
Mahesh SathiamoorthyMS#1005|@MADIATOR
Chris PaxtonCP#732|@CHRIS_J_PAXTON
Robert ScobleRS#321|@SCOBLEIZER
will depueWD#254|@WILLDEPUE
swyxSW#214|@SWYX
240

Shift offers free apartment cleaning in NYC in exchange for video data to train domestic robotsVetted operators wear cameras to record physical workflow data.

13h|760.8k2.1k645
PrakashPR#1332|@8TEAPI
signüllSI#1124|@SIGNULLL
Mahesh SathiamoorthyMS#1005|@MADIATOR
Chris PaxtonCP#732|@CHRIS_J_PAXTON
Robert ScobleRS#321|@SCOBLEIZER
will depueWD#254|@WILLDEPUE
swyxSW#214|@SWYX
250

Thomas Bloom and collaborators disprove the sum-product conjecture, sparking debate over AI's role in the breakthroughThe work was inspired by OpenAI's unit distance counterexample.

20h|651.4k8.6k1.6k
Patrick ShaftoPS#1426|@PATRICKSHAFTO
bilalBI#1402|@BILALTWOVEC
Boaz BarakBB#133|@BOAZBARAKTCS
Noam BrownNB#31|@POLYNOAMIAL
260

Trump administration issues grand jury subpoenas to X and Reddit to unmask anonymous ICE criticsThe legal fight over user privacy remains under seal.

17h|1.8M9.4k2.1k
270

Liquid AI releases LFM2.5-8B-A1B, an open-weight, device-optimized Mixture-of-Experts model with 1.5 billion active parametersThe 128K context model was trained on 38 trillion tokens.

14h|330.2k2.6k1.7k
merveME#675|@MERVENOYANN
Joscha BachJB#671|@PLINZ
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(#420|@TEORTAXESTEX
clem 🤗C🤗#68|@CLEMENTDELANGUE
AKAK#28|@_AKHALIQ
280

A16z partner Justine Moore posts a parody about using a hypothetical "Claude Opus 4.8" model to rename a single fileThe joke satirizes using massive AI models for trivial tasks.

12h|1.8M29.3k2.3k
fofrFO#1996|@FOFRAI
Matthew BermanMB#1759|@MATTHEWBERMAN
Justine MooreJM#931|@VENTURETWINS
Mohammad NorouziMN#467|@MO_NOROUZI
290

Tesla's Ashok Elluswamy says Cybercab will soon begin autonomous Austin driving after factory exit demoThe vehicle exited Giga Texas without a human occupant.

10h|#1VIEWS32M#1LIKED200.7k7.6k
Ashok ElluswamyAE#1489|@AELLUSWAMY
Elon MuskEM#76|@ELONMUSK
300

Tesla's Ashok Elluswamy teases Cybercab autonomous capabilities with footage of the vehicle driving inside Giga TexasPublic road testing in Austin is expected to begin soon.

10h|32M200.7k7.6k
Ashok ElluswamyAE#1489|@AELLUSWAMY
Elon MuskEM#76|@ELONMUSK