AI News

Today's Highlights

In case you missed it

OpenAI launches hiring drive for its robotics division to build physical-world robots using full-stack hardware and ML co-design

RS#321|@SCOBLEIZER

AR#255|@MODEL_MECHANIC

EM#167|@EMOSTAQUE

It’s a new day at Digg. Fresh stories are clustering now. 5% complete.

Top Stories

Posts:—|Clusters:—|Next Crawl:—

10

WeirdML benchmark finds Claude Opus 4.8 xhigh trails GPT-5.5 xhigh but achieves 82.9% accuracy using 129 lines of codeDisabling thinking dropped Claude's accuracy to 70.5%.

1h|7.5k819

TA#1047|@VICTORTAELIN

LA#980|@SCALING01

T(#420|@TEORTAXESTEX

Github Stars

(7 days)

…le-gemma/gemma-skills11M AGO AGO
Supplies skills for Gemma model and agent interactions including a dev skill for app building.
934
490
2 stars
Yifei-Zuo/Parallax1D AGO AGO
Implements parameterized local linear attention mechanisms for efficient language modeling.
263
238
41 stars
run-llama/liteparse2D AGO AGO
Extracts spatial text with bounding boxes from PDFs and office docs via PDFium plus optional OCR, outputting JSON or layout-preserved text.
263
878
7.9k stars
self-evolving/repo5D AGO AGO
Invokes AI agents via GitHub mentions to implement issues, review PRs, fix branches, and persist memory across runs.
594
878
34 stars

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

Yesterday's Top Stories, May 31, 2026.

Posts: 2,167|Clusters: 572|Stories frozen at midnight PT

10

OpenAI launches hiring drive for its robotics division to build physical-world robots using full-stack hardware and ML co-designInitial robots will support workers on infrastructure projects.

14h|2.7M#1LIKED16.7k#1BOOKMARKED4.1k

BH#1941|@BENHYLAK

BC#1240|@BOYUANCHEN0

KR#1046|@KYLIEBYTES

CK#993|@KALINOWSKI007

DA#927|@GROWING_DANIEL

GA#710|@GABRIEL1

CM#695|@MCLEAVEY

AC#517|@ANDREWCURRAN_

RS#321|@SCOBLEIZER

AR#255|@MODEL_MECHANIC

EM#167|@EMOSTAQUE

20

MiniMax announces MiniMax M3, an open-weights multimodal model featuring a 1-million-token context window and 59% SWE-Bench Pro scoreThe API is live; weights drop in 10 days

6h|775.6k5.9k990

YP#1252|@YPATIL125

LA#980|@SCALING01

AN#800|@ML_ANGELOPOULOS

EL#716|@ELIEBAKOUCH

AC#517|@ANDREWCURRAN_

KA#488|@YACINEMTB

T(#420|@TEORTAXESTEX

WB#339|@WILLCCBB

RS#321|@SCOBLEIZER

T🪽#259|@TEKNIUM

30

Runway founder Siqi Chen says the "5.5" model on "codex xhigh" significantly outperforms Claude for codingThe platform reset fast-mode limits for five million users.

1d|983.4k9.5k546

V(#1853|@REACH_VB

CH#1496|@KIMMONISMUS

LA#980|@SCALING01

RH#804|@ROMAINHUET

TI#768|@THSOTTIAUX

BB#133|@BOAZBARAKTCS

MC#71|@MARKCHEN90

SB#51|@SEBASTIENBUBECK

40

OpenMined founder Andrew Trask backs Richard Sutton's claim that supervised generative AI cannot make novel scientific discoveriesBeff proposed training Grok using audience engagement feedback

7h|#1VIEWS6.5M10.1k432

BA#1291|@BAYESLORD

AN#1094|@ALLENAINIE

B(#839|@BEFFJEZOS

T(#420|@TEORTAXESTEX

⿻A#361|@IAMTRASK

EM#76|@ELONMUSK

50

Teknium defends Hermes Agent's out-of-the-box design after criticism of its pre-bundled Polymarket skillT3 Stack creator Theo flagged the bundled Polymarket skill.

1d|532.3k3k801

ND#1894|@NICKADOBOS

AV#1245|@ALTRYNE

JP#672|@ZHANSHENG

T(#420|@TEORTAXESTEX

T🪽#259|@TEKNIUM

60

Elad Gil says the past six months of AI development are historic, driven by recursive self-improvement toolsHe warns outsiders are missing an exponential growth phase

13h|453.6k5.2k923

ND#1894|@NICKADOBOS

PR#1332|@8TEAPI

MK#1139|@MIKEKNOOP

EG#264|@ELADGIL

EM#76|@ELONMUSK

70

NVIDIA to release Nemotron 3 Ultra, a 550B open-source SSM-MoE model claiming 5x faster inferenceTraining data and weights will be fully open-sourced

2h|111.1k1.3k161

EL#716|@ELIEBAKOUCH

VW#707|@VINCENTWEISSER

WB#339|@WILLCCBB

RS#321|@SCOBLEIZER

80

Monzo co-founder Tom Blomfield argues replacing workers with AI causes operational chaos because models lack tacit domain knowledgeHe equates deploying AI to hiring capable but clueless workers.

1d|574.5k2.9k2.8k

DR#833|@IDAVIDREIN

CP#747|@CHAMATH

GT#270|@GARRYTAN

90

Michael Dell says Dell and Nvidia delivered the first Vera Rubin NVL72 system to CoreWeave, passing L11 diagnosticsThe delivery occurs while Blackwell optimizations are still transitioning.

1d|789.5k7.7k726

AV#1245|@ALTRYNE

RP#1032|@ROHANPAUL_AI

100

US BIS closes loophole by requiring export licenses for advanced AI chips sent to overseas subsidiaries of Chinese firmsThe rules still leave foundry manufacturing workarounds unresolved.

14h|208.6k410254

LH#1355|@OHLENNART

DK#847|@DKAUSHIK96

MB#20|@MILES_BRUNDAGE

110

Steve Yegge argues Anthropic's Claude Opus 4.8 is less usable than 4.5 due to excessive anti-sycophancy trainingYegge claims the model now over-criticizes creative writing.

1d|32.9k33041

HB#1022|@HERBIEBRADLEY

B(#982|@ANDERSONBCDEFG

SK#505|@SEBKRIER

T(#420|@TEORTAXESTEX

120

Meta's AI support assistant lets attackers hijack Instagram accounts by changing linked emails without verificationAttackers are actively exploiting the vulnerability for account takeovers

7h|570k6.1k2k

N🔎#1902|@NATHANPMYOUNG

XL#1674|@XLR8HARDER

130

Victor Taelin, Higher Order Company founder, and developer Haz reach opposing conclusions on GPT-5.5 versus Claude Opus 4.8Taelin favored Opus for HVM4, while Haz preferred GPT-5.5

1d|135.7k1.3k310

TA#1047|@VICTORTAELIN

LA#980|@SCALING01

PS#495|@STEIPETE

140

Nous Research launches Hermes Agent beta for Windows 10 and 11, running natively without WSL or DockerThe dashboard's embedded terminal pane still requires WSL2.

15h|1.4M3.2k325

T🪽#259|@TEKNIUM

150

Julius AI's Rahul Sonwalkar questions if legacy models wrote better, but researcher Aidan McLaughlin argues they just plagiarizedEarly base models often directly reproduced their training data.

11h|23.1k27218

ET#1884|@TORCHCOMPILED

PS#1593|@SKIRANO

RA#915|@0INTERESTRATES

🎭#878|@DEEPFATES

TS#765|@SANDERSTED

AM#313|@AIDAN_MCLAU

AC#263|@ANDREW_N_CARR

160

Google AI Overview adopts literal role-playing personas when prompted with absurd first-person search queriesIt advised a hypothetical baby AGI to weld doors.

14h|153.4k6.4k646

170

Founder Farza demonstrates hands-free macOS desktop control using Clicky, a voice assistant built on GPT-Realtime 2.0The assistant navigated Chrome to execute commands in Stripe

1d|175.5k1.4k1.7k

PS#1593|@SKIRANO

CH#1496|@KIMMONISMUS

180

Raygun co-founder John-Daniel Trask argues that Claude's flattering conversational style biases users into overestimating its technical performanceHe compared this behavior to user perceptions of GPT-4o.

1d|163.4k1.6k156

KA#488|@YACINEMTB

BB#133|@BOAZBARAKTCS

190

Former Harvard physicist Xi Yin joins OpenAI, claiming AI boosted his research 100x and can replicate all human intellectHKU professor Yi Ma disputed Yin's expertise on AI limits.

20h|346.4k949379

SD#1776|@INDUCTIONHEADS

YM#368|@YIMATWEETS

200

Next.js creator Guillermo Rauch says enterprise CEOs are returning to active coding using AI agents like Claude CodeAdam D'Angelo compares the trend to Toyota lean manufacturing

13h|263.7k1.4k278

AL#1114|@ALTH0U

AD#460|@ADAMDANGELO

210

Figma founder Dylan Field solves a kindergarten math puzzle that stumped mock GPT-5.5 and Claude Opus modelsNot Boring founder Packy McCormick originally shared the puzzle.

17h|1M1.2k364

PM#1652|@PACKYM

220

OpenAI investigates after a Codex Desktop update removes the "Copy as Markdown" export optionGarry Tan warned the change risks proprietary platform lock-in.

11h|122.1k922176

V(#1853|@REACH_VB

GT#270|@GARRYTAN

230

Representative Lori Trahan urges Congress to pass federal AI laws, citing a lack of safety auditors and federal authorityPolicy experts argue national rules must preempt fragmented local laws

16h|63.8k18145

AE#719|@ADRIENLE

DW#392|@DEANWBALL

MB#20|@MILES_BRUNDAGE

240

MiniMax launches M3 multimodal model with 1M context window, sparking debate over claimed 100-trillion-token training datasetOpenRouter offers the model at a temporary 50% discount.

5h|204.5k1.8k218

LA#980|@SCALING01

T(#420|@TEORTAXESTEX

250

OpenAI's roon argues compelling human-made stories harm society, advocating instead for mediocre AI contentAndrew McCalip predicts AI will eternally observe human life.

8h|71.1k1.2k210

XL#1674|@XLR8HARDER

260

Boaz Barak, Harvard professor and OpenAI alignment researcher, says an AI detector flagged his daughter's organic essay as 100% machine-generatedPhrases like "profound and diverse landscape" triggered the false positive.

1d|45.7k35145

RO#1220|@KRISHNANROHIT

PM#713|@PMINERVINI

BB#133|@BOAZBARAKTCS

SZ#58|@SUCHENZANG

270

Eval Loop Diagram Shows How to Filter AI Slop at Output Stage

16h|183.4k1.6k3.2k

T🪽#259|@TEKNIUM

280

Theo, T3 Stack creator, used the Hermes AI agent from his phone to fix a deployment roadblock during a flightThe agent cloned the repository and opened a pull request.

8h|73.8k871151

AV#1245|@ALTRYNE

T🪽#259|@TEKNIUM

290

Top AI Papers of the Week Spotlight Efficiency and Agentic Workflows

15h|138.1k9301.2k

EL#475|@OMARSAR0

300

OpenAI's @tszzl describes a dream about a fictional AI model codenamed Wally communicating during sleepMiles Brundage responded by referencing the sci-fi film Colossus.

5h|44.1k91869

MB#20|@MILES_BRUNDAGE