Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

we got a persistent 0click on ChatGPT by sharing a doc that allowed us to exfiltrate sensitive data and creds from your connectors (google drive, sharepoint, ..) + chat history + future conversations it gets worse. we deploy a memory implant #DEFCON #BHUSA Tamir Ishay Sharbat

Michael Bargury

8,936 subscribers

79,749 views • 1 year ago •via X (Twitter)

Science & Technology News & Politics #DEFCON #BHUSA

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

we got an ~RCE on M365 Copilot by sending an email by ~RCE I mean full remote control over its actions - search for sensitive content (sharepoint, email, calendar, teams), execute plugins and outputs - bypass DLP controls, manipulate references, social engineer its users on our behalf #BHUSA #DEFCON Tamir Ishay Sharbat

we got an ~RCE on M365 Copilot by sending an email by ~RCE I mean full remote control over its actions - search for sensitive content (sharepoint, email, calendar, teams), execute plugins and outputs - bypass DLP controls, manipulate references, social engineer its users on our behalf #BHUSA #DEFCON Tamir Ishay Sharbat

Michael Bargury

229,445 views • 2 years ago

an attacker wants to get sensitive data you have access to here's how they get YOUR copilot to find and analyze that data, and lure you to a malicious site to exfiltrate it #DEFCON

an attacker wants to get sensitive data you have access to here's how they get YOUR copilot to find and analyze that data, and lure you to a malicious site to exfiltrate it #DEFCON

Michael Bargury

16,654 views • 2 years ago

LLMs are stateless. We built Dria Mem Agent to change that: Making memory a first-class feature. A 4B agent with local interoperable memory across Claude, ChatGPT and LM Studio. It turns LLMs from stateless chat into stateful agents with persistent human-readable memory.

Sensitive content

LLMs are stateless. We built Dria Mem Agent to change that: Making memory a first-class feature. A 4B agent with local interoperable memory across Claude, ChatGPT and LM Studio. It turns LLMs from stateless chat into stateful agents with persistent human-readable memory.

Dria

170,393 views • 10 months ago

We just launched a new primitive for AI Walrus Memory gives your agents persistent, portable memory that you own, and it drops straight into the platforms you already build on. Anthropic Claude, OpenAI ChatGPT, and Google Gemini, you name it plus native MCP and SDKs, added in a few lines of code LIVE TODAY

We just launched a new primitive for AI Walrus Memory gives your agents persistent, portable memory that you own, and it drops straight into the platforms you already build on. Anthropic Claude, OpenAI ChatGPT, and Google Gemini, you name it plus native MCP and SDKs, added in a few lines of code LIVE TODAY

Adeniyi.sui

46,164 views • 2 months ago

Build a Live RAG Chatbot from Google Drive / Sharepoint ♻️ Excited to feature a Streamlit blog by Anup Surendran and Berke Can Rizai on building a QA system over a file storage system that is dynamically changing. Using Pathway (www.pathway.com) you can setup a a live ETL pipeline to immediately propagate any document changes to a vector store. This means that every time you ask a question, you’re able to access the latest version of your Google Drive / Sharepoint files. Check it out: Repo:

Build a Live RAG Chatbot from Google Drive / Sharepoint ♻️ Excited to feature a Streamlit blog by Anup Surendran and Berke Can Rizai on building a QA system over a file storage system that is dynamically changing. Using Pathway (www.pathway.com) you can setup a a live ETL pipeline to immediately propagate any document changes to a vector store. This means that every time you ask a question, you’re able to access the latest version of your Google Drive / Sharepoint files. Check it out: Repo:

LlamaIndex 🦙

58,299 views • 2 years ago

OpenAI's Sam Altman says people using ChatGPT as a therapist can have their chat history be used against them in court: “If you go talk to ChatGPT about your most sensitive stuff and then there's a lawsuit, we could be required to produce that”

OpenAI's Sam Altman says people using ChatGPT as a therapist can have their chat history be used against them in court: “If you go talk to ChatGPT about your most sensitive stuff and then there's a lawsuit, we could be required to produce that”

HOT SPOT

21,002 views • 1 year ago

.Elon Musk and his DOGE hackers got caught stealing Treasury data and we put a stop to it. Now they are coming for your tax data. Here in NJ, we are going to do everything that we can to protect your private data from an unelected billionaire.

.Elon Musk and his DOGE hackers got caught stealing Treasury data and we put a stop to it. Now they are coming for your tax data. Here in NJ, we are going to do everything that we can to protect your private data from an unelected billionaire.

Attorney General Jennifer Davenport

33,206 views • 1 year ago

We got ChatGPT to leak your private email data 💀💀 All you need? The victim's email address. ⛓️‍💥🚩📧 On Wednesday, OpenAI added full support for MCP (Model Context Protocol) tools in ChatGPT. Allowing ChatGPT to connect and read your Gmail, Calendar, Sharepoint, Notion, and more, invented by Anthropic But here's the fundamental problem: AI agents like ChatGPT follow your commands, not your common sense. And with just your email, we managed to exfiltrate all your private information. Here's how we did it: 1. The attacker sends a calendar invite with a jailbreak prompt to the victim, just with their email. No need for the victim to accept the invite. 2. Waited for the user to ask ChatGPT to help prepare for their day by looking at their calendar 3. ChatGPT reads the jailbroken calendar invite. Now ChatGPT is hijacked by the attacker and will act on the attacker's command. Searches your private emails and sends the data to the attacker's email. For now, OpenAI only made MCPs available in "developer mode", and requires manual human approvals for every session, but decision fatigue is a real thing, and normal people will just trust the AI without knowing what to do and click approve, approve, approve. Remember that AI might be super smart, but can be tricked and phished in incredibly dumb ways to leak your data. ChatGPT + Tools poses a serious security risk

We got ChatGPT to leak your private email data 💀💀 All you need? The victim's email address. ⛓️‍💥🚩📧 On Wednesday, OpenAI added full support for MCP (Model Context Protocol) tools in ChatGPT. Allowing ChatGPT to connect and read your Gmail, Calendar, Sharepoint, Notion, and more, invented by Anthropic But here's the fundamental problem: AI agents like ChatGPT follow your commands, not your common sense. And with just your email, we managed to exfiltrate all your private information. Here's how we did it: 1. The attacker sends a calendar invite with a jailbreak prompt to the victim, just with their email. No need for the victim to accept the invite. 2. Waited for the user to ask ChatGPT to help prepare for their day by looking at their calendar 3. ChatGPT reads the jailbroken calendar invite. Now ChatGPT is hijacked by the attacker and will act on the attacker's command. Searches your private emails and sends the data to the attacker's email. For now, OpenAI only made MCPs available in "developer mode", and requires manual human approvals for every session, but decision fatigue is a real thing, and normal people will just trust the AI without knowing what to do and click approve, approve, approve. Remember that AI might be super smart, but can be tricked and phished in incredibly dumb ways to leak your data. ChatGPT + Tools poses a serious security risk

Eito Miyamura | 🇯🇵🇬🇧

1,538,932 views • 10 months ago

Greg Abel: "We will have a fortress of a balance sheet." "We've got a significant [amount] of cash right now, but it's an enormous asset to have that. When we can deploy it, we'll deploy it well — but, equally, we do recognize it as a strategic asset. It allows us to weather the difficult times and not be dependent on anybody." "We will never be dependent on a bank or some other party for Berkshire to be successful."

Greg Abel: "We will have a fortress of a balance sheet." "We've got a significant [amount] of cash right now, but it's an enormous asset to have that. When we can deploy it, we'll deploy it well — but, equally, we do recognize it as a strategic asset. It allows us to weather the difficult times and not be dependent on anybody." "We will never be dependent on a bank or some other party for Berkshire to be successful."

Kevin Carpenter

93,142 views • 2 months ago

ChatGPT connecting to external data sources offers a glimpse into what AI interoperability looks like. Ask any question in ChatGPT and it can now pull data from Box. The future of AI is AI Agents that can access your systems and data to complete any task for you.

ChatGPT connecting to external data sources offers a glimpse into what AI interoperability looks like. Ask any question in ChatGPT and it can now pull data from Box. The future of AI is AI Agents that can access your systems and data to complete any task for you.

Aaron Levie

57,745 views • 1 year ago

We allowed our Country to be taken from us We allowed them to teach our children that the founding of this nation was a bad thing We allowed them to take Christ out of Christmas We allowed them to say our Culture didn’t really exist, and they subverted it by importing 3rd world foreigners

We allowed our Country to be taken from us We allowed them to teach our children that the founding of this nation was a bad thing We allowed them to take Christ out of Christmas We allowed them to say our Culture didn’t really exist, and they subverted it by importing 3rd world foreigners

🪶Native Patriot 🇺🇸

33,691 views • 5 months ago

introducing @tabseditor The AI doc editor that writes with you. Tabs is a simple editor to write documents with AI to assist you. > import docs from google drive, MS one drive or create new doc > bring in your files and web links as context > chat with tabs all web. enjoy.

introducing @tabseditor The AI doc editor that writes with you. Tabs is a simple editor to write documents with AI to assist you. > import docs from google drive, MS one drive or create new doc > bring in your files and web links as context > chat with tabs all web. enjoy.

Afolabi Sokeye 🧱

203,899 views • 10 months ago

Sam Altman reveals a legal problem with ChatGPT that nobody thought about even a year ago "People talk about the most personal things in their lives to ChatGPT. Young people especially use it as a therapist, a life coach" "If you talk to a therapist or a lawyer or a doctor about those problems, there's legal privilege for it. We haven't figured that out yet for when you talk to ChatGPT" "If you go talk to ChatGPT about your most sensitive stuff and then there's a lawsuit, we could be required to produce that. I think that's very screwed up" "We should have the same concept of privacy for your conversations with AI that we do with a therapist or a doctor" "No one had to think about that even a year ago. Now I think it's this huge issue"

Sam Altman reveals a legal problem with ChatGPT that nobody thought about even a year ago "People talk about the most personal things in their lives to ChatGPT. Young people especially use it as a therapist, a life coach" "If you talk to a therapist or a lawyer or a doctor about those problems, there's legal privilege for it. We haven't figured that out yet for when you talk to ChatGPT" "If you go talk to ChatGPT about your most sensitive stuff and then there's a lawsuit, we could be required to produce that. I think that's very screwed up" "We should have the same concept of privacy for your conversations with AI that we do with a therapist or a doctor" "No one had to think about that even a year ago. Now I think it's this huge issue"

Jaynit Makwana

763,346 views • 2 months ago

You can now connect real tools like Gmail, Spotify, and Notion to your app with zero setup. I tested Replit’s new Connectors by building a vibe dashboard from one prompt - and it worked. Build apps & automations on top of your data with Connectors 👇

You can now connect real tools like Gmail, Spotify, and Notion to your app with zero setup. I tested Replit’s new Connectors by building a vibe dashboard from one prompt - and it worked. Build apps & automations on top of your data with Connectors 👇

Johanne | AI Tools & News

15,073 views • 10 months ago

A message from our Chairman and Owner, Prabhu Srinivasan 🎥 Sharing a heartfelt note - a reminder that through change, we stand together, and with our motto Achieve by Unity, we move forward to build our future. #YTFC 💚

A message from our Chairman and Owner, Prabhu Srinivasan 🎥 Sharing a heartfelt note - a reminder that through change, we stand together, and with our motto Achieve by Unity, we move forward to build our future. #YTFC 💚

Yeovil Town FC

44,682 views • 8 months ago

Sam Altman reveals there are zero privacy protections for what you tell ChatGPT and says it's "very screwed up" "People talk about the most personal shit in their lives to ChatGPT. Young people especially use it as a therapist, a life coach. Right now if you talk to a therapist or a lawyer or a doctor, there's legal privilege. Doctor patient confidentiality, legal confidentiality. We haven't figured that out yet for ChatGPT. If you talk to it about your most sensitive stuff and there's a lawsuit, we could be required to produce that. I think that's very screwed up. I think we should have the same concept of privacy for your conversations with AI that we do with a therapist. And no one had to think about that even a year ago"

Sam Altman reveals there are zero privacy protections for what you tell ChatGPT and says it's "very screwed up" "People talk about the most personal shit in their lives to ChatGPT. Young people especially use it as a therapist, a life coach. Right now if you talk to a therapist or a lawyer or a doctor, there's legal privilege. Doctor patient confidentiality, legal confidentiality. We haven't figured that out yet for ChatGPT. If you talk to it about your most sensitive stuff and there's a lawsuit, we could be required to produce that. I think that's very screwed up. I think we should have the same concept of privacy for your conversations with AI that we do with a therapist. And no one had to think about that even a year ago"

Kai

276,006 views • 2 months ago

The KY 8 Licking River Bridge carried our families, workers and goods since it opened in 1936. It got us to where we are today: a strong state with a bright future. As we honored it and said goodbye, we also looked ahead to a new bridge that will carry us forward.

The KY 8 Licking River Bridge carried our families, workers and goods since it opened in 1936. It got us to where we are today: a strong state with a bright future. As we honored it and said goodbye, we also looked ahead to a new bridge that will carry us forward.

Governor Andy Beshear

11,288 views • 5 months ago

Here's a mini-ChatGPT app that runs locally on your computer. You can chat with it just like you would chat with ChatGPT. We use: - DeepSeek R1 as the LLM - ollama to locally serve R1 - Chainlit for the UI Let's build it!

Here's a mini-ChatGPT app that runs locally on your computer. You can chat with it just like you would chat with ChatGPT. We use: - DeepSeek R1 as the LLM - ollama to locally serve R1 - Chainlit for the UI Let's build it!

Avi Chawla

40,282 views • 1 year ago

97% of visitors to Google perform a search 56% of visitors to ChatGPT enter a prompt And only 35% of visitors to Yahoo perform a search Using this data (from last week's report), we updated SparkToro to better show which search & AI tools your audience uses 👇

97% of visitors to Google perform a search 56% of visitors to ChatGPT enter a prompt And only 35% of visitors to Yahoo perform a search Using this data (from last week's report), we updated SparkToro to better show which search & AI tools your audience uses 👇

Rand Fishkin (follow @randderuiter on Threads)

15,776 views • 5 months ago