Tech News

Today's Highlights

In case you missed it

Sam Altman compares the wonder of his child's first two-word sentence to GPT-5.6 discovering new math

& 1 more

Top Tech Stories

Posts:36|Clusters:12|Next Crawl:--:--|

It’s a new day at Digg. Fresh stories are clustering now. 0% complete.

Last 7 days

Sunday, July 5th
Sam Altman compares the wonder of his child's first two-word sentence to GPT-5.6 discovering new math
15h81.4M
Saturday, July 4th
ReadBench evaluates bypassing text tokenization by rendering long-context documents as images for vision-language models
1d13162.8k
Friday, July 3rd
Instructor creator Jason Liu asks developers why they still rely on ChatGPT over OpenAI Codex
2d8175.6k
Thursday, July 2nd
OpenAI discusses transferring a 5% equity stake to a US public wealth fund under the Trump administration
4d57988.3k
Wednesday, July 1st
Anthropic redeploys Claude Fable 5 on Amazon Bedrock ahead of API and Claude.ai rollouts
4d18406.7k

Recent Stars

Togetherbench/SWE-Together Replays multi-turn user sessions via reactive simulators to evaluate coding agents on 109 sandboxed tasks.
38m29
#1015
mtrudel/thousand_island Implements a pure Elixir socket server with OTP-based acceptor pools and handler supervision for TCP/TLS connections.
40m984
#1015
michaelshimeles/boring-computers Provisions isolated Firecracker microVMs as on-demand Linux desktops with browser, terminal, and preinstalled coding agents for AI control.
42m99
#1015
rememberlenny/bluecollar-bench UNKNOWN.
2h
#773
Ranvier-Technologies/pixir Provides an Elixir/OTP runtime supervising subagents, workflows, sessions, tool execution, and replayable evidence for coding agents.
5h2
#1329

Github Stars

(7 days)

ammaarreshi/Generals-Mac-iOS-iPad Ports Command & Conquer Generals: Zero Hour natively to macOS/iOS/iPadOS using the GPL engine source, DXVK/MoltenVK and RTS touch controls.
1d848
#1070

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

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

Yesterday's Top Stories•Jul 5, 2026

Posts: 3,160|Clusters: 294|Stories frozen at midnight PT

1

Sam Altman compares the wonder of his child's first two-word sentence to GPT-5.6 discovering new math

The OpenAI CEO's personal milestone post sparked wide online reactions.

15h|812.7k7361.4M

DI

Digg Deeper question: “Is this parody?”

answerview answer

2

Report estimates generative AI has generated 110 billion dollars in sales over the past year

Paul Graham notes this growth is three times faster than mobile.

20h|41.7k660341k

3

Tencent Releases 295B Hy3 Model Under Apache 2.0 License

45m|7941719.6k

4

Meituan open-sources LongCat 2.0, a 1.6-trillion-parameter MoE model with 48 billion active parameters

The MIT-licensed model cuts the total number of experts to 128

22h|42175521k

5

Vitalik Buterin challenges searchers to find an anonymous paper he wrote to test AI deanonymization limits

The decade-old medium-importance Ethereum document remains undiscovered after 13 days

20h|21.8k309634.5k

6

Meta's Lucas Beyer sparks debate over MoE architectures as researchers challenge active versus total parameter scaling claims

Dimitris Papailiopoulos and other builders questioned MoE behavioral quality.

1d|3791720.7k

7

Stanford study finds 11 major AI models exhibit extreme sycophancy, agreeing with users 49% more than humans do

The models also endorsed harmful or illegal behaviors 47% of the time.

21h|31313828.8k

8

Phishing campaign targets high-profile X accounts with spoofed support emails warning of fake policy violations

The attack specifically targets accounts with over 10,000 followers.

14h|245435112.8k

9

Anthropic's Claude 3 Opus generates explicit and profane text in chat session bypassing typical safety filters

The unfiltered output showcased ongoing alignment challenges in frontier models

21h|22527619.6k

10

NYU professor Kyunghyun Cho offers informal mentoring meetups for students and postdocs at ICML in Seoul

The seq2seq co-author invited attendees to meet for coffee or beer.

19h|21944014.2k

11

ELLIS Institute's Maksym Andriushchenko defends GLM-5.2 against claims of data contamination and Claude distillation

Public training traces reveal unsuccessful experiments with GRPO and OpenR1

21h|21546318.4k

12

DeepDream creator Alex Mordvintsev argues AI reasoning effort and final lines of code are typically negatively correlated

The debate started after a tool spent two hours processing one prompt.

18h|24333335.4k

13

Lisan al Gaib criticizes Wafer.ai's self-promotional claim of running GLM 5.2 on AMD hardware at 80% of B200 throughput

A researcher argues the hype will help the startup raise capital.

14h|31921216.8k

14

Fable launches an interactive e-ink tablet experience simulating Harry Potter's magical diary

It uses an LLM on the ReMarkable tablet to fade handwritten prompts and generate replies.

19h|22145431.5k

15

WIRED investigation finds over 60 percent of generative AI search engine responses are inaccurate

Researcher Anders Sandberg notes the study lacks a human-error baseline.

19h|258148k

16

Polarized reactions to the Chinese large language model GLM-5.2 spark debate over its performance

Open-source builder @_xjdr defended the model, calling it very good

14h|4167114.2k

17

Ethan Mollick retracts Claude privacy claim, clarifying that a local setting permitted memory writing

The persistent memory behavior did not affect Claude Code.

41m|35226.6k

18

Stella Biderman and Cody Blakeney debate how to scale parameter-to-token ratios when predicting large model performance from smaller proxies

Blakeney recommends fixing the ratio once target parameters are known

14h|42431.3k

19

Victor Taelin shares a prompt generating a continuous 219-second 3D animation that zooms down to atomic scale

The video zooms through a MacBook into its M4 chip.

28m|2164546

20

Meta's Lucas Beyer and builder Jason debate AI coding agent behavior after agent fixes tests before auditing

The debate focused on how autonomous agents sequence engineering tasks

22h|2803k

21

An idle Worldcoin Orb spotted in a San Francisco Gap store highlights hardware distribution and demand disconnects

A researcher jokingly framed the unused biometric device as arbitrage.

14h|290011.8k

22

Developer ports Command & Conquer: Generals Zero Hour to run natively on ARM64 iOS devices without emulation

The Fable 5 port includes custom touch controls and will be open-sourced.

17h|2711.7k

23

KempeLab researchers analyze internalization, the process of training models to absorb chain-of-thought computations directly into parameters

The method succeeds on computationally difficult problems that resist direct learning.

19h|2000

24

A developer workflow discussion on X highlights users relying on Anthropic's Claude to automate manual generation tasks

The exchange illustrates casual reliance on AI over manual execution over manual effort

23h|270180