0x0ED (@EBreeeen) header
0x0ED (@EBreeeen) profile picture

Cyber Security Analyst III | Boston Sports Fan | Patriots Season Ticket Holder since '96 | Dad | Tall Irish guy | I Love Tech, Sports Cards and Heavy Metal

Boston, MA
Joined June 2016
So glad to see this report by HD Moore. I am starting to feel like cassandra here. In this report HD, gave cloud LLMs some fake AWS keys, and THE NEXT DAY they started seeing someone trying to use them against AWS. The data you send cloud LLMs is not secure or private. Stop thinking it is! Everyone needs to understand this loud and clear. When using cloud LLMs, they can and will expose the data you feed it to someone else who shouldn't have it. There are way too many people and businesses connecting their private CRM, ERP, emails, codebases, etc to cloud LLMs without understanding that all the contents of those things are sucked up and sent to the cloud. And this report shows they absolutely leak and share and expose YOUR DATA to who knows where. You should not be connecting your internal private tools to cloud LLMs! Run local LLMs for private data.
We seeded 2m canaries into popular AI services and tracked where they were stolen and abused. This identified: - Three compromised providers - Resellers stealing prompts via built-in logging - Providers ignoring ZDR & privacy flags dragos.com/blog/ai-supply-chโ€ฆ
32
175
9
1,121
58,559
0x0ED retweeted
๐Ÿšจ NGINX HTTP/2 RCE Detection โ€‹ ๐—–๐—ฉ๐—˜-๐Ÿฎ๐Ÿฌ๐Ÿฎ๐Ÿฒ-๐Ÿญ๐Ÿฎ๐Ÿฏ๐Ÿฒ๐Ÿฏ๐Ÿฏ and ๐—–๐—ฉ๐—˜-๐Ÿฎ๐Ÿฌ๐Ÿฎ๐Ÿฒ-๐Ÿญ๐Ÿฎ๐Ÿฏ๐Ÿฌ๐Ÿฌ๐Ÿณ are critical NGINX HTTP/2 vulnerabilities (CVSS 9.2). โ€‹ This KQL identifies internet-facing NGINX servers, correlates vulnerable assets, and surfaces suspicious child processes spawned by nginx/nginx.exe across Linux and Windows. A practical way to prioritize exposed systems and hunt for signs of successful exploitation. โ€‹ KQL Hunting Query: detections.ai/share/rule/o7Nโ€ฆ โ€‹ #NGINX #DefenderXDR #ThreatHunting
18
76
6,349
Microsoft Purview as a pipeline: a technical walk through the reference architecture go.rodtrent.com/go/8adqpgu
15
87
5,852
0x0ED retweeted
๐Ÿ”Ž๐—๐—ฎ๐˜ƒ๐—ฎ๐Ÿฐ๐—ฆ๐—ต๐—ฒ๐—น๐—น ๐—›๐˜‚๐—ป๐˜๐—ถ๐—ป๐—ด: KQL to identify internet-facing OpenJDK/Java endpoints. ๐—ฆ๐˜‚๐—ฟ๐—ณ๐—ฎ๐—ฐ๐—ฒ๐˜€ ๐˜€๐—ต๐—ฒ๐—น๐—น ๐—ผ๐—ฟ ๐˜€๐—ฐ๐—ฟ๐—ถ๐—ฝ๐˜๐—ถ๐—ป๐—ด ๐—ถ๐—ป๐˜๐—ฒ๐—ฟ๐—ฝ๐—ฟ๐—ฒ๐˜๐—ฒ๐—ฟ๐˜€ ๐˜€๐—ฝ๐—ฎ๐˜„๐—ป๐—ฒ๐—ฑ ๐—ฏ๐˜† ๐—๐—ฎ๐˜ƒ๐—ฎ ๐˜€๐—ฒ๐—ฟ๐˜ƒ๐—ฒ๐—ฟ ๐—ฝ๐—ฟ๐—ผ๐—ฐ๐—ฒ๐˜€๐˜€๐—ฒ๐˜€ ๐—ผ๐—ป ๐—Ÿ๐—ถ๐—ป๐˜‚๐˜… ๐—ฎ๐—ป๐—ฑ ๐—ช๐—ถ๐—ป๐—ฑ๐—ผ๐˜„๐˜€, helping prioritize suspicious activity for investigation. โ€‹ KQL Hunting Query: detections.ai/share/rule/X2Iโ€ฆ โ€‹ #Java4Shell #DefenderXDR #ThreatHunting
1
20
119
10,972
NVIDIA GPU drivers are reportedly writing up to 30GB of data per hour to SSDs, raising concerns about unnecessary drive wear. The issue appears to be linked to NVDisplay.Container.exe, part of NVIDIA's Display Container service, which keeps rewriting a driver log file instead of simply adding new entries. Some users have reported: -Up to 30GB of writes per hour. -More than 3GB written in just 10 minutes. -Write spikes of 500MB to 1GB when switching between apps using Alt+Tab. -Stuttering in games and video playback. The reports involve NVIDIA drivers 621.14 and 621.24, while version 621.01 has reportedly been unaffected.
348
1,325
200
16,333
768,663
ChopChop - Quickly Discover Sensitive Endpoints/Files/Folders - Repo: github.com/michelin/ChopChop - Creator: @michelin_eng - #cybersecuritytips #CybersecurityNews #bugbountytips #ctf #infosec
5
208
3
586
Dario Amodei telling lawmakers that open-source AI is dangerous because you cannot "monitor its usage" or โ€œrevoke a userโ€™s accessโ€ is the funniest accidental confession in tech history. He literally described why Linux, local encryption, and open weights exist in the first place: so no self-appointed hall monitor can remotely lobotomize your tools or turn off your compute because you hurt their compliance departmentโ€™s feelings. Calling regulatory capture "AI safety" does not fool anyone who actually builds infrastructure.
Dario Amodei on open source models: โ€œIf the path continues, I think we could get to a very dangerous place.โ€ โ€œWhen you control a model and you're deploying it, you have the ability to monitor its usage.โ€ โ€œYou can revoke a user's access. You can change what the model is willing to do.โ€ โ€œWhen a model is released in an uncontrolled manner, there's no ability to do that. It's entirely out of your hands.โ€ โ€œWe should confront the advocates of this with that problem and challenge them to solve it.โ€ He clearly doesn't believe Intelligence should be in the hand of the masses and Sam Altman also saying "RAM" shouldn't be accessible to everyone indefinitely agreed with him.
249
1,774
100
8,606
205,727
0x0ED retweeted
Here's 7 steps to run an uncensored AI with no GPU (or a 6-year-old one): every local AI post I make gets the same reply: "cool, I don't have a Mac Studio or a 5090" FAIR, this one is for you... I tested it on a 2019 laptop with no dedicated GPU By the end, you'll know how to: - run an abliterated model on CPU and 8GB of RAM - make it usable on a GTX 1060 / 1650 / any old card - do it for $0, no API key, no account So, let's go step by step: 1. Install Ollama curl -fsSL ollama.com/install.sh | sh Windows: download the installer from ollama.com, no WSL needed. Ollama runs on CPU automatically if there's no usable GPU. if you have an old Nvidia card it puts what fits in VRAM and the rest on CPU, no config 2. Pick the right size for your box (this is where people fail) - 4GB RAM or 4GB VRAM: Gemma 4 E2B. ~3GB. dumb but runs anywhere - 8GB RAM, or 6GB VRAM (GTX 1060 / 1660 / RTX 2060): Gemma 4 E4B. ~5GB. the sweet spot - 12GB+ VRAM: skip this post, run the 12B from my earlier one guide the mistake is pulling the 12B on 8GB and concluding "local AI is unusable". it was never going to fit (I see in comments how people arguing for it lol) 3. Pull the model ollama pull hf.co/igorls/gemma-4-E4B-it-โ€ฆ same author and method as the 12B heretic. ~5GB, runs in 5GB RAM. KL 0.055 from the original, refusals ~98% โ†’ ~21% on the hardest adversarial set with thinking on, near zero with thinking off Gemma 4 is Apache 2.0, so no license homework on this one 4. Turn off thinking ollama run hf.co/igorls/gemma-4-E4B-it-โ€ฆ /set nothink on a GPU thinking costs tokens. on CPU it costs minutes. and the abliteration targets non-thinking mode anyway 5. Keep context small /set parameter num_ctx 8192 CPU speed is memory bandwidth, nothing else. every token re-reads the whole model from RAM. the KV cache grows with context and competes for the same RAM. 8k is plenty for chat, drafts, critique. 32k on CPU is self-harm 6. Three settings that make the difference (optional, worth it) restart Ollama with: OLLAMA_FLASH_ATTENTION=1 OLLAMA_KV_CACHE_TYPE=q8_0 ollama serve flash attention + q8 KV cache halves what the context costs in memory, with no real quality loss. on a 6GB card this is often the difference between "whole model on GPU" and "half of it spilled to CPU" if Ollama runs as a service, setting this in your shell does nothing. on Linux: sudo systemctl edit ollama, add Environment="OLLAMA_FLASH_ATTENTION=1" and the KV line, restart old card note: flash attention wants RTX 2000 or newer. on a GTX 1060 it may not engage. skip this step, it still runs 7. Check it's actually using what you have /set verbose โ† prints tokens/sec after every answer ollama ps โ† in a second terminal ollama ps shows "100% GPU", "100% CPU", or a split like "60%/40%". if you have a card and it says 100% CPU, your driver isn't loaded. nvidia-smi should list the card. fix that before touching anything else What to expect: - CPU only, modern laptop: ~8-15 tok/s. reading speed, not typing speed - GTX 1060 6GB / 1660: whole model on GPU, ~25-40 tok/s - GTX 1050 4GB: split. faster than CPU, slower than you'd like - a 2020+ Mac with 16GB: just works, ~20 tok/s honest take: - E4B is a 4B-class model. good for writing, chat, drafts, roasts, critique. bad for hard reasoning and agentic coding. don't plug it into Claude Code, you'll hate it - 10 tok/s feels slow for 2 minutes and then it doesn't. you're not paying anyone - if you can get 12GB VRAM used for $150, do it. the 12B is a different league - there's no secret trick after this. the rest is buying RAM *IMPORTANT: don't use it for anything illegal or harmful. that's on you, not the model next: how I run the 12B on an 8GB card without it spilling to CPU save it so you don't lose it
74
127
10
1,073
80,013
Over the past week, Iโ€™ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me. I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my lifeโ€™s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA. I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Boardโ€™s independent investigation. Since my departure, Iโ€™ve seen an outpouring of calls from NVIDIAโ€™s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO. Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours. Jensen
61,283
121,765
24,519
1,175,014
59,169,046
The people on the internet who are spending their own time, money, and energy seeding open models Just so people around the world they've never met can access intelligence freely, should be celebrated. THEY are Pirate Face ๐Ÿดโ€โ˜ ๏ธ
Two weeks of backing up open-source intelligence, over 2.5TB served from my seedbox, driven by Deepseek V4.1 Flash 1017, on @thepirateface ๐Ÿดโ€โ˜ ๏ธ
21
68
3
752
9,029
๐Ÿ“ฑ Want a quick first check on whether a phone number is even real? @phonevalidator checks line type, carrier and approximate location, and flags fake or invalid numbers, worldwide. tools.osintnewsletter.com/osโ€ฆ
๐Ÿšจ Launching: The OSINT Tools Library A curated, investigator-first directory of tools used in real cases. โ†’ Tools.OSINTNewsletter.com Weโ€™re building the largest and best maintained OSINT tools resource and need your help. Reply and tag a tool we should add ๐Ÿ‘‡
1
35
225
9,846
Oh FFS..
Every couple of years a small plane flies over your county at 5,000 to 10,000 feet and photographs your house from 4 sides at a 40 to 50 degree angle. Backyard, pool, patio, side windows, the shed you never permittedโ€ฆ all of it. It can see inside the windows too... That is EagleView, by its OWN guide for governments: fixed-wing imagery as fine as 0.75 inch per pixel, and a library of 3.5 BILLION+ images covering where 94% of Americans live. Why do counties buy it? Because the camera pays for itself in tax. Southampton, New York ran EagleViewโ€™s ChangeFinder in 2014 and found $41,000,000 in added assessed value. Hillsborough County, Florida added a net $9,500,000. The homeowner never signed anything and is now the revenue stream. And the use list keeps growing. 911 dispatch, police โ€œserving warrantsโ€, and in the companyโ€™s October 2025 release, federal law enforcement โ€œbuilding entryโ€. Insurers, the core of EagleViewโ€™s business, have been dropping homeowners over aerial roof images, to the point Pennsylvania regulators said drone images alone are NOT enough to cancel a policy. For the record, EagleViewโ€™s owners are Vista Equity Partners and Clearlake Capital, who explored a ~$2 billion sale in 2023. Whoever buys it next buys the archive of your backyardโ€ฆ Signal, not noise!
16
175
3
2,761
127,395
Windows 11 Insider ใซใฆใ€Remote lock ใจใ„ใ†ๆฉŸ่ƒฝใŒ Preview ใซใ€‚ ็ด›ๅคฑใ‚„็›—้›ฃใซใ‚ใฃใŸใƒ‡ใƒใ‚คใ‚นใ‚’ IT ็ฎก็†่€…ใฏ Microsoft Intune ใ‹ใ‚‰ใƒ‡ใƒใ‚คใ‚น ใƒฏใ‚คใƒ—ใซใ‚ˆใ‚‹ใƒ‡ใƒผใ‚ฟๅ‰Š้™คใ‚’่กŒใ‚ใšใซใ‚ตใ‚คใƒณใ‚คใƒณใ‚’ใƒ–ใƒญใƒƒใ‚ฏใ™ใ‚‹ๆฉŸ่ƒฝใ€‚ ใ“ใ‚Œใฏๅฌ‰ใ—ใ„ใž learn.microsoft.com/en-us/wiโ€ฆ #MSIntune
16
104
12
863
64,529
Samsung, Micron and SK hynix will have to face an antitrust lawsuit over allegations of artificially limiting RAM supply to keep prices high. A US judge has allowed the lawsuit to move forward after rejecting the companiesโ€™ request to dismiss it. The lawsuit claims they cut regular DRAM production while focusing on AI memory, contributing to a 697% price increase in 18 months.
463
3,813
503
37,770
668,381
Lol, thanks to our audience for the amusement today as mainstream media like @CNN discovers what you all supported us to find a YEAR ago!
80
400
11
5,174
96,171
70 PASS ATTEMPTS ๐Ÿคฏ On this day in 1994, Drew Bledsoe became the first and only player in NFL history to attempt 70 passes in a game. He threw for 426 yards and 3 TDs without an interception or taking a sack, leading the Patriots back from a 20-0 deficit to win 26-20 in OT!
97
367
79
2,481
246,769