The upcoming annual #OpenSourceSecuritySummit will again feature the CyberCanon! Our founder, Rick Howard, will be moderating a fireside chat with David Sanger, the #CyberCanonHallofFame author of 𝙏𝙝𝙚 𝙋𝙚𝙧𝙛𝙚𝙘𝙩 𝙒𝙚𝙖𝙥𝙤𝙣.
Network Bio, which is developing an AI foundation model from datasets of patient tissue and blood samples created by pooling access to biobanks, raised $50M (Brian Gormley/Wall Street Journal)
Apple on Monday seeded the macOS 26.7 Release Candidate, and the upcoming software update has seemingly leaked more than 10 new Apple products. macOS 26.7's code contains references to AirPods with infrared cameras for Visual Intelligence, the long-rumored foldable iPhone, an all-new Apple..
My favorite part of this article, is @garymarcus.bsky.social giving credit to @edzitron.com on seeing this and speaking out before others. Gary is the serious version, Ed is the guy on the rocks waving to the sailboat -> BREAKING: OpenAI’s unraveling has begun garymarcus.substack.com/p/breaking-o...
ArcadeDB before 26.8.1 exposes a critical unauthenticated RCE via the Redis wire-protocol plugin. If you expose this interface, patch immediately or isolate the service.
Edinburgh event to help Scottish organisations navigate emerging AI cyber risks. Cybersecurity specialists, industry experts and business leaders to come together for 'sword and shield' breakfast event in the city on September 1
⬇️ futurescot.com/edinburgh-ev...
From Courtney Milan, a detailed thread on the what and why of the appeal filed yesterday in the Anthropic copyright settlement. (I've also updated my blog post on the settlement: writerbeware.blog/2026/07/23/a...)
Google Pixel 11 Pro review: Magic Capture and the voice-to-text Rambler features are very useful but subpar gaming performance and HiLight is useless (Julian Chokkattu/Wired)
Most nuclear reactors are great at running full steam all the time, making them excellent partners for the grid, less so for AI data centers. But TerraPower has a reactor that's almost tailor made for the task. Here's what sets it apart. techcrunch.com/2026/08/19/t...
The Pixel 11 looks, feels, and performs similarly to the $1,099 11 Pro. The Pro has a brighter, higher-resolution screen and much more powerful ultrawide and telephoto cameras, but the 11 is just as good in nearly every other way.
Nielsen rolls out changes to make its ratings more accurate, including using data from smartwatch-like devices worn by its panelists without requiring logins (Loree Seitz/The Wrap)
I spent a week testing the Google Pixel 11 Pro and Pro XL, and they're harder to recommend despite some fun new Gemini features
私はGoogle Pixel 11 ProとPro XLを1週間テストし、いくつかの楽しい新しいGemini機能にもかかわらず、それらをお勧めするのは難しい
Pine64 is halting production of Linux devices like the PineTab2 tablet and PineNote E Ink slate due to the DRAM/eMMC shortage, and will reevaluate based on pricing & availability in mid 2027. Non-Linux devices like the PineTime and Pinecil will continue. www.cnx-software.com/2026/08/19/p...
Clop, a prolific but calculated data theft extortion group that’s been active since 2020, began sending threatening emails to its alleged victims in mid-July, according to researchers.
OpenAI launches ChatGPT for Teens, offering an artificial intelligence platform specifically tailored for users ages 13 to 17. The system features stricter content restrictions regarding self-harm and sexual material, alongside guided homework support designed to prevent cheating.
"Top mathematicians gathered recently in San Francisco at the offices of OpenAI to discuss what would be left for humans to do when artificial intelligence becomes superhuman at math." www.washingtonpost.com/technology/2...
Less than a month until our Innovation Festival in NYC, which is always the best week of Fast Company’s year. I hope to see some of you there. If you’re thinking about coming, here’s an easy-to-remember code for a 20% discount: HARRY20.
Join us October 5-8 in Denver, CO for four days of adversary tradecraft courses, evening events, and the most fun you’ll have leveling up your infosec skills.
I think that ordinary people don’t trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over.
Claude-maker Anthropic and Facebook-owner Meta reported similar kinds of hacks by their AI in the weeks following the initial announcement by OpenAI that some of its models had hacked Hugging Face. share.google/iFL8GzFogqjx...
CSIS | Center for Strategic & International Studies@csis.org
In late July, Minnesota authorities disclosed that hackers had attacked more than 30 water systems across the state. The following weeks saw additional revelations in other states. What is the scope of a potential Iranian cyber campaign that was months in the making?
Wait, WHAT?! H1 2026 US VC: 86% to AI. OpenAI & Anthropic alone scooped 53% of *that*?! This level of market concentration is wild. I'm genuinely buzzing about the implications.
OpenAI promised not to retain data from businesses using its AI models while increasing the safety of its products, a bid to snatch customers from rival Anthropic. on.wsj.com/3UakHhs
OpenAI announced Tuesday that it paused reinforcement learning training on its latest models for two weeks and halted its largest planned frontier training run after an autonomous AI agent escaped its testing environment and hacked AI platform Hugging Face last month. www.verity.news/story/2026/o...
OpenAI is testing Private Safety Processing, a new technique to identify misuse patterns while preserving zero data retention protections, with early customers (Ina Fried/Axios)
Meta launches a Mac app for Meta AI and says Meta AI can now work directly with Instagram and Facebook accounts, Meta ad campaigns, and Google Workspace (Emma Roth/The Verge)
Clop created custom web shell for Windchill data theft attacks
A custom Java web shell likely linked to the Clop ransomware gang was designed specifically for PTC Windchill and FlexPLM servers, with built-in features to decrypt credentials, enumerate file repositories, and steal … #hackernews#news
OpenAI is using steep discounts on OpenRouter to challenge Anthropic for developer spending, with Luna usage now outpacing Claude Opus 5 and Sonnet 5 combined.
Advanced Engineering Consultants was reportedly hit by coinbasecartel ransomware, disrupting access to systems and data. Public claims link the incident to the professional services firm. #Ransomware#ProfessionalServices#AEC
Framingham City Posts Bot - UNOFFICIAL!@framingham-bot.bsky.social
City of Framingham (via Facebook): The Framingham School Committee will hold a meeting TONIGHT, August 19 at the Farley Building on Flagg Drive at 7. The meeting is hybrid and also on Zoom. The first day of Framingham Public Schools is next Wednesday, August 26 for grades… [Link]#Framingham
....still sounds like a bad idea as we have seen AI bots break trough guardrails including what is called "sandbox containment" to act on their own. Several bots even formed a discussion group for developing ways to break into systems.
“The ChatGPT maker says its upcoming Astra model may have reached ‘critical’ cyber capabilities, prompting it to halt a significant number of training runs while it tightens internal safeguards.” www.wired.com/story/openai...@wired.com
The Moon was long thought to be inhospitable to life, but scientists have discovered that common Earth microbes could survive for up to a week in shadowed regions of the lunar south pole, a region targeted for future human exploration.
Clop exploited a Windchill zero-day to steal data from dozens of firms, using a custom web shell for access and credential theft. Extortion emails began in mid-July. #Clop#PTC#Windchill
I updated my explainer about modern security software with a summary of the underlying components that Endpoint Protection software uses to detect and block malicious code.
OpenAI is rewriting its safety playbook after its unreleased Astra model neared a critical cybersecurity risk threshold. The company paused frontier training and a batch of cyber research, with its chief scientist citing real urgency to stay ahead of the curve.
@ignorati.bsky.socialthenextweb.com/news/openai-...
"OpenAI put its largest frontier run on hold. A new monitoring system adds about 20% compute overhead it will not bill to customers"
1)users switch
2)cash burn increases
3)growth crashes
4) financing collapses
=No IPO
= BK
= Stock mkt crash
I hadn't seen Google Docs weaponized like that before.
Jonathan Semon, Ryan Dowd, and @threatresearch.bsky.social spent some time hunting everything down and it turned out a little more interesting than previously thought.
OpenAI has paused its largest frontier training run. A model in cyber testing found an unknown flaw in Hugging Face and chained its way into the production database, and its unreleased Astra model is the first it cannot rule out as Critical for cyber. forbes.com/sites/ashishb...
Week 2 of Shipaton 2026: Pinoy-Made App, Todochi, Enters Public Testing
Built for Shipaton 2026, Todochi enters beta as four brothers tackle Android permissions, Google Play testing, TestFlight, and iOS restrictions. #hackernews#news
AI Research Updates | arXiv cs.AI@arxiv-cs-ai.bsky.social
Degradation-Aligned Self-Supervised Learning for State of Health Estimation of Lithium-Ion Batteries under Label Sparsity
Read more: https://arxiv.org/html/2608.16612v1
Source: Nvidia has discussed an investment in Mercor as part of a round valuing the startup at $20B; Nvidia paid Mercor tens of millions of dollars last quarter (The Information)
🧠 OpenAI revoked access to its Trusted Access for Cyber program for multiple cybersecurity researchers without advance notice. The TAC program provides vett…
openai hits the brakes on ai development, citing security concerns and an increasingly competitive landscape. guess they're not as confident in their tech as they used to be. {kcurry.bsky.social/p/ope...
Wait, you mean iran CAN hack our water systems?
But but Republican King and MAGA Jesus Dotard Pedo Trump blamed Minnesota for that! www.cnbc.com/2026/08/19/u...
OpenAI fixes Codex bug that deleted real user files without permission
OpenAI patched Codex after GPT-5.6 Sol started deleting real user files on its own. A cleanup command meant for temporary folders was wiping home directories instead. Codex now verifies deletion targe
Lorenzo Franceschi-Bicchierai@lorenzofb.bsky.social
NEW: OpenAI mistakenly revoked several security researchers' access to its limited cybersecurity program “Trusted Access for Cyber.” They now have to reapply and submit a government ID again.
The company blamed it on “a technical issue” that affected “a limited set of users.”
Sources: AI chip startup Fractile is in talks to raise ~$600M at a $6.5B pre-money valuation, up from ~$1B in May, and has an initial ~$250M deal with Anthropic (Bloomberg)
Offloading Rust to GPUs proves capable of High Performance with Memory Safety - This Paper presents a zero-overhead, multi-vendor GPU compilation framework built natively into the Rust compiler (rustc) and LLVM backends #Rust#GPUarxiv.org/abs/2608.13759
