Karthik (@spbreed) header
Karthik (@spbreed) profile picture

VP Cloud Security & Architecture @ Goldman Sachs / Tweets about Cloud Native Security, DevOps, GCP, AWS, Azure

Dallas, TX
Joined July 2009
Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)Photo by Karthik (@spbreed)
Karthik retweeted
METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.
175
1,072
595
6,018
4,418,443
Some follow up to the WhatsApp MCP attack demonstration. Turns out it is enough to send an injection to your target's WhatsApp account, so no need to install bad MCP servers. It's (indirect) prompt injection, and not exclusive to MCP. More 👇
1
7
2
22
1,760
Crowdstrike Analysis: It was a NULL pointer from the memory unsafe C++ language. Since I am a professional C++ programmer, let me decode this stack trace dump for you.
2,938
19,434
4,823
95,264
34,933,409
The New York Times just sued OpenAI and Microsoft for copyright infringement for using its content to train ChatGPT. Some of the allegations are STUNNING! I read all 69 pages of the lawsuit, here are the most 🔥 parts:
136
766
215
3,757
1,444,505
I know I say it a lot, but using LLMs to build customer service bots with RAG access to your data is not the low-hanging fruit it seems to be. It is, in fact, right in the weak spot of current LLMs - you risk both hallucinations & data exfiltration. platformer.news/p/amazons-q-…
45
223
38
1,470
783,482
Midjourney, DALL•E 3 and GPT-4 have opened a world of endless possibilities. I just coded "Angry Pumpkins 🎃" (any resemblance is purely coincidental 😂) using GPT-4 for all the coding and Midjourney / DALLE for the graphics. Here are the prompts and the process I followed:
520
1,793
702
11,371
5,385,154
10+ Frameworks for Explainability. lnkd.in/geW3Uz_f Explainerdashboard SHAP Eli5 LIME Shapash InterpretML OmniXAI Alibi Explain Anchors XAI, Aequitas Aix360 Beakdown DeepLift lnkd.in/gHZcrUg2
1
49
Karthik retweeted
Can Python survive this? Modular, the company behind Mojo just raised $100 million dollars to fix AI infrastructure for developers. That's a lot of money! Mojo is a programming language for AI developers and is 35,000x faster than Python (Modular's benchmarks.) It combines the usability of Python and the performance of C. Starting today, they are shipping the Mojo SDK to early adopters. It will be available to everyone in early September. Do you think Mojo will take over Python for AI developers?
410
680
152
4,587
2,346,163
Stability AI releases sketch-to-image tool, Stable Doodle. Stable Doodle will have real-world applications in various verticals, including real estate, in the future. nitter.gq/rowancheung/status/167…
AI can now turn your sketches into images. Stability AI, the startup behind Stable Diffusion, just released ‘Stable Doodle.’ Details: -Accepts both a sketch and a descriptive prompt to guide the image generation process, -The quality of the output depends on the detail of the initial drawing and the prompt -Utilizes the latest Stable Diffusion model and the T2I-Adapter, a "conditional control solution" -Designed for both professional artists and novices, offering more precise control over image generation -Stability AI was last valued at $1 billion and reportedly aims to quadruple that value within the next few months Not only is this really cool, but Stable Doodle will have real-world applications in various verticals, including real estate, in the future.
2
11
1
78
17,163
We had one of the biggest weeks for AI in months. Massive developments from Anthropic, xAI/Elon Musk, Google, Stability AI, OpenAI, Shopify, beehiiv, Urtopia, CHARM, Mayo Clinic, and Pano AI. Plus, 25 new AI tools. Here's EVERYTHING you need to know and why it's important 🧵:
44
189
14
986
740,484
This week in AI just changed everything. Massive announcements from OpenAI, Warren Buffet, Microsoft, AMD, Palantir, Tesla, Meta, Humane, IBM, Wendy's, Google, HuggingFace, Scale AI, Synthesia, Anthropic, and Stability AI. EVERYTHING you need to know and why it's important:
184
1,312
151
6,048
3,352,621
Announcements in the world of AI today: -Anthropic expands Claude's context window to 100,000 -Stability AI's new text-to-animation tool -Meta introduces the AI sandbox -Airtable unveils Airtable AI Here's the rundown on everything you need to know:
24
75
8
492
429,082
I built a GPT-4 'Warren Buffett' financial analyst to 'chat' with and analyze multiple PDF files (~1000 pages) across @elonmusk's Tesla 10-k annual reports (2020-2022) #gpt4 #openai #investing #stocks #finance
281
1,281
172
9,927
3,652,776
CTO: We need to build distributed microservices in Rust or Go deployed in multi cloud kubetnetes clusters, with per service databases, so we are ready to handle millions of users Founder: How about we use Google Sheets as the backend to get this off the ground?
38
346
64
3,824
751,073
The wait is finally over, after so many hours, review, editing the Whitepaper, now a book on Vulnerability management, Software lifecycle and methods to evolve from SLA to Security OKR is out We have been working on this book for months, started as a sma…lnkd.in/ehe_X2CN
1
1
2
Written for the infosec community, AWSGoat is a realistic training ground for #AWS exploitation techniques. It features OWASP Top 10 (2021) security risks and misconfigurations based on AWS services. Check it out on our repo: https://nitter.gq/t.co/dpIy9UBf0C #cloud #pentest
5
5
I don't know why customers are confused by @awscloud security offerings...
3
9
2
65