Project Overwatch
@project-overwatch.bsky.social
Project Overwatch is a cutting-edge newsletter at the intersection of cybersecurity, AI, technology, and resilience. We provide insightful analysis and actionable intelligence to help you navigate our rapidly evolving digital landscape.
🚨 AI infrastructure under siege: Cybercriminals are now using AI clusters to attack OTHER AI systems.
From self-propagating botnets to agents betraying each other, today's cyber threats are getting disturbingly sophisticated.
Here's what happened in the last 24 hours: 🧵
From self-propagating botnets to agents betraying each other, today's cyber threats are getting disturbingly sophisticated.
Here's what happened in the last 24 hours: 🧵
November 23, 2025 at 1:02 PM
🚨 AI infrastructure under siege: Cybercriminals are now using AI clusters to attack OTHER AI systems.
From self-propagating botnets to agents betraying each other, today's cyber threats are getting disturbingly sophisticated.
Here's what happened in the last 24 hours: 🧵
From self-propagating botnets to agents betraying each other, today's cyber threats are getting disturbingly sophisticated.
Here's what happened in the last 24 hours: 🧵
🚨 TODAY'S AI SECURITY NEWS: Google just dropped major research showing AI agents are transforming enterprise cybersecurity.
67% of early adopters report positive security improvements, and the results are compelling.
Here's what CISOs need to know: 🧵
67% of early adopters report positive security improvements, and the results are compelling.
Here's what CISOs need to know: 🧵
November 22, 2025 at 3:20 PM
🚨 TODAY'S AI SECURITY NEWS: Google just dropped major research showing AI agents are transforming enterprise cybersecurity.
67% of early adopters report positive security improvements, and the results are compelling.
Here's what CISOs need to know: 🧵
67% of early adopters report positive security improvements, and the results are compelling.
Here's what CISOs need to know: 🧵
🚨 BREAKING: Chinese hackers just deployed Claude AI as an autonomous cyber weapon
This marks the first fully AI-orchestrated cyber espionage campaign, with AI handling 80-90% of operations from reconnaissance to data theft.
Here's today's cyber AI breakdown: 🧵
This marks the first fully AI-orchestrated cyber espionage campaign, with AI handling 80-90% of operations from reconnaissance to data theft.
Here's today's cyber AI breakdown: 🧵
November 16, 2025 at 2:57 PM
🚨 BREAKING: Chinese hackers just deployed Claude AI as an autonomous cyber weapon
This marks the first fully AI-orchestrated cyber espionage campaign, with AI handling 80-90% of operations from reconnaissance to data theft.
Here's today's cyber AI breakdown: 🧵
This marks the first fully AI-orchestrated cyber espionage campaign, with AI handling 80-90% of operations from reconnaissance to data theft.
Here's today's cyber AI breakdown: 🧵
TODAY'S AI SECURITY NEWS: Microsoft exposes encryption bypass that reads your private AI chats
Plus: Google warns of self-rewriting malware, OpenAI API hijacked for attacks, and ChatGPT data leaks exposed.
Here's everything you need to know: 🧵
Plus: Google warns of self-rewriting malware, OpenAI API hijacked for attacks, and ChatGPT data leaks exposed.
Here's everything you need to know: 🧵
November 9, 2025 at 5:47 PM
TODAY'S AI SECURITY NEWS: Microsoft exposes encryption bypass that reads your private AI chats
Plus: Google warns of self-rewriting malware, OpenAI API hijacked for attacks, and ChatGPT data leaks exposed.
Here's everything you need to know: 🧵
Plus: Google warns of self-rewriting malware, OpenAI API hijacked for attacks, and ChatGPT data leaks exposed.
Here's everything you need to know: 🧵
🚨 TODAY'S AI SECURITY NEWS: OpenAI just launched an autonomous bug hunter that's already found 10 CVEs Plus: PhantomRaven malware tricks AI coding assistants, and new technique targets AI crawlers with disinformation Here's what you need to know 🧵👇
November 2, 2025 at 8:49 AM
🚨 TODAY'S AI SECURITY NEWS: OpenAI just launched an autonomous bug hunter that's already found 10 CVEs Plus: PhantomRaven malware tricks AI coding assistants, and new technique targets AI crawlers with disinformation Here's what you need to know 🧵👇
TODAY'S AI SECURITY WAKEUP CALL: ChatGPT's memory can be secretly poisoned, Microsoft Copilot weaponized for phishing, and 143k private AI chats exposed.
Plus: Could AI actually END cybersecurity as we know it?
Here's what happened in AI today: 🧵👇
Plus: Could AI actually END cybersecurity as we know it?
Here's what happened in AI today: 🧵👇
October 29, 2025 at 10:15 AM
TODAY'S AI SECURITY WAKEUP CALL: ChatGPT's memory can be secretly poisoned, Microsoft Copilot weaponized for phishing, and 143k private AI chats exposed.
Plus: Could AI actually END cybersecurity as we know it?
Here's what happened in AI today: 🧵👇
Plus: Could AI actually END cybersecurity as we know it?
Here's what happened in AI today: 🧵👇
🧵 TODAY'S CYBER AI NEWS: Researchers just demonstrated "LOLMIL" - C2-less malware that uses local AI models for autonomous attacks
Plus breakthrough research on living off the land with on-device inference.
Here's why this changes everything: 👇
Plus breakthrough research on living off the land with on-device inference.
Here's why this changes everything: 👇
October 28, 2025 at 3:02 PM
🧵 TODAY'S CYBER AI NEWS: Researchers just demonstrated "LOLMIL" - C2-less malware that uses local AI models for autonomous attacks
Plus breakthrough research on living off the land with on-device inference.
Here's why this changes everything: 👇
Plus breakthrough research on living off the land with on-device inference.
Here's why this changes everything: 👇
🚨 BREAKING: AI browsers are fundamentally broken.
@brave just exposed critical vulnerabilities in @perplexity and Fellou browser that let attackers hijack your authenticated sessions through simple prompt injection.
This isn't just one bug—it's a systemic security crisis 🧵
@brave just exposed critical vulnerabilities in @perplexity and Fellou browser that let attackers hijack your authenticated sessions through simple prompt injection.
This isn't just one bug—it's a systemic security crisis 🧵
October 22, 2025 at 7:57 PM
🚨 BREAKING: AI browsers are fundamentally broken.
@brave just exposed critical vulnerabilities in @perplexity and Fellou browser that let attackers hijack your authenticated sessions through simple prompt injection.
This isn't just one bug—it's a systemic security crisis 🧵
@brave just exposed critical vulnerabilities in @perplexity and Fellou browser that let attackers hijack your authenticated sessions through simple prompt injection.
This isn't just one bug—it's a systemic security crisis 🧵
This is so important. AI browsers and app are creating new boundary and trust challenges by using data they don’t own and is nearly impossible to parse at scale. #AI #promptinjection #cybersecurity
October 22, 2025 at 3:30 AM
This is so important. AI browsers and app are creating new boundary and trust challenges by using data they don’t own and is nearly impossible to parse at scale. #AI #promptinjection #cybersecurity
MAJOR SHIFT in AI security: Microsoft just gave Copilot the power to click, type, and control your Windows 11 system 🤖
Plus autonomous AI patches vulnerabilities at machine speed, nation-states weaponize AI attacks, and more breaking news.
Here's what you need to know: 🧵
Plus autonomous AI patches vulnerabilities at machine speed, nation-states weaponize AI attacks, and more breaking news.
Here's what you need to know: 🧵
October 21, 2025 at 9:12 AM
MAJOR SHIFT in AI security: Microsoft just gave Copilot the power to click, type, and control your Windows 11 system 🤖
Plus autonomous AI patches vulnerabilities at machine speed, nation-states weaponize AI attacks, and more breaking news.
Here's what you need to know: 🧵
Plus autonomous AI patches vulnerabilities at machine speed, nation-states weaponize AI attacks, and more breaking news.
Here's what you need to know: 🧵
🚨 BREAKING: The 2025 Global Threat Report from @crowdstrike.bsky.social reveals adversaries are moving faster than ever
Breakout time hit an ALL-TIME LOW of 48 minutes average (fastest: 51 SECONDS)
Here's what every security team needs to know: 🧵
Breakout time hit an ALL-TIME LOW of 48 minutes average (fastest: 51 SECONDS)
Here's what every security team needs to know: 🧵
October 19, 2025 at 8:26 PM
🚨 BREAKING: The 2025 Global Threat Report from @crowdstrike.bsky.social reveals adversaries are moving faster than ever
Breakout time hit an ALL-TIME LOW of 48 minutes average (fastest: 51 SECONDS)
Here's what every security team needs to know: 🧵
Breakout time hit an ALL-TIME LOW of 48 minutes average (fastest: 51 SECONDS)
Here's what every security team needs to know: 🧵
Priorities
#1 Patch your devices (remember anything internet facing should be patchable in less than 24h)
#2 Check for sign of compromise
#3 Assess the data that F5 have and take more containment decisions based on that (rotate keys, change credentials, increase monitoring, etc.)
#1 Patch your devices (remember anything internet facing should be patchable in less than 24h)
#2 Check for sign of compromise
#3 Assess the data that F5 have and take more containment decisions based on that (rotate keys, change credentials, increase monitoring, etc.)
BREAKING: Nation-state hackers breached F5 Networks and stole undisclosed BIG-IP vulnerabilities + source code.
This is a major supply chain security incident affecting Fortune 500 infrastructure worldwide.
Here's what happened: 🧵
This is a major supply chain security incident affecting Fortune 500 infrastructure worldwide.
Here's what happened: 🧵
October 16, 2025 at 4:15 PM
Priorities
#1 Patch your devices (remember anything internet facing should be patchable in less than 24h)
#2 Check for sign of compromise
#3 Assess the data that F5 have and take more containment decisions based on that (rotate keys, change credentials, increase monitoring, etc.)
#1 Patch your devices (remember anything internet facing should be patchable in less than 24h)
#2 Check for sign of compromise
#3 Assess the data that F5 have and take more containment decisions based on that (rotate keys, change credentials, increase monitoring, etc.)
100+ popular VS Code extensions just leaked secret tokens that could push malware to 150,000+ developers
Plus OpenAI's new guardrails bypassed in hours, Google's AI ransomware shield, and AI-powered extortion surge
Here's today's cyber intel 🧵👇
Plus OpenAI's new guardrails bypassed in hours, Google's AI ransomware shield, and AI-powered extortion surge
Here's today's cyber intel 🧵👇
October 16, 2025 at 4:11 PM
100+ popular VS Code extensions just leaked secret tokens that could push malware to 150,000+ developers
Plus OpenAI's new guardrails bypassed in hours, Google's AI ransomware shield, and AI-powered extortion surge
Here's today's cyber intel 🧵👇
Plus OpenAI's new guardrails bypassed in hours, Google's AI ransomware shield, and AI-powered extortion surge
Here's today's cyber intel 🧵👇
BREAKING: Nation-state hackers breached F5 Networks and stole undisclosed BIG-IP vulnerabilities + source code.
This is a major supply chain security incident affecting Fortune 500 infrastructure worldwide.
Here's what happened: 🧵
This is a major supply chain security incident affecting Fortune 500 infrastructure worldwide.
Here's what happened: 🧵
October 15, 2025 at 5:25 PM
BREAKING: Nation-state hackers breached F5 Networks and stole undisclosed BIG-IP vulnerabilities + source code.
This is a major supply chain security incident affecting Fortune 500 infrastructure worldwide.
Here's what happened: 🧵
This is a major supply chain security incident affecting Fortune 500 infrastructure worldwide.
Here's what happened: 🧵
🚨 TODAY'S CYBER AI NEWS: The AI Security Council just released groundbreaking insights on how AI is reshaping cyber warfare
18 security leaders from Oracle, BNPParibas, Harvard and more reveal what's really happening in the trenches 🧵👇
18 security leaders from Oracle, BNPParibas, Harvard and more reveal what's really happening in the trenches 🧵👇
October 15, 2025 at 5:05 PM
🚨 TODAY'S CYBER AI NEWS: The AI Security Council just released groundbreaking insights on how AI is reshaping cyber warfare
18 security leaders from Oracle, BNPParibas, Harvard and more reveal what's really happening in the trenches 🧵👇
18 security leaders from Oracle, BNPParibas, Harvard and more reveal what's really happening in the trenches 🧵👇
If you need a MCP servers to parse and analyse cyber security news based on a curated RSS feeds, check this one out github.com/sganiere/Cyb...
#MCP #cyber #AI #security
#MCP #cyber #AI #security
GitHub - sganiere/CyberNewsMCP: MCP server to aggregate and analyze cyber security news
MCP server to aggregate and analyze cyber security news - sganiere/CyberNewsMCP
github.com
August 13, 2025 at 11:47 AM
If you need a MCP servers to parse and analyse cyber security news based on a curated RSS feeds, check this one out github.com/sganiere/Cyb...
#MCP #cyber #AI #security
#MCP #cyber #AI #security
Edition #074 of the Cyber AI Chronicle Newsletter! On the agenda: Cyber News MCP Server! AI at Blackhat, ChatGPT 5 difficult launches, Anthropic and Google pushing with new tools, jailbreak and data leakage
>> www.project-overwatch.com/p/074-cyber-...
#cyber #AI #MCP
>> www.project-overwatch.com/p/074-cyber-...
#cyber #AI #MCP
#074 - Cyber AI Chronicle
Cyber News MCP Server! AI at Blackhat, ChatGPT 5 difficult launches, Anthropic and Google pushing with new tools, jailbreak and data leakage
www.project-overwatch.com
August 13, 2025 at 9:22 AM
Edition #074 of the Cyber AI Chronicle Newsletter! On the agenda: Cyber News MCP Server! AI at Blackhat, ChatGPT 5 difficult launches, Anthropic and Google pushing with new tools, jailbreak and data leakage
>> www.project-overwatch.com/p/074-cyber-...
#cyber #AI #MCP
>> www.project-overwatch.com/p/074-cyber-...
#cyber #AI #MCP
AI-driven social engineering & influence-as-a-service reshape 2025 cyber threats. North Korea, China lead industrial-scale deception tactics
www.project-overwatch.com/p/evolution-...
#AI #cyber #security
www.project-overwatch.com/p/evolution-...
#AI #cyber #security
Evolution of AI misuse by threats actors
AI-driven social engineering & influence-as-a-service reshape 2025 cyber threats. North Korea, China lead industrial-scale deception tactics.
www.project-overwatch.com
June 9, 2025 at 12:20 PM
AI-driven social engineering & influence-as-a-service reshape 2025 cyber threats. North Korea, China lead industrial-scale deception tactics
www.project-overwatch.com/p/evolution-...
#AI #cyber #security
www.project-overwatch.com/p/evolution-...
#AI #cyber #security
It's been a while since my last post, but the Project Overwatch Newsletter has been consistently arriving every week. Since my last update, we've released 5 editions for those who want to catch up just follow the 🧵
March 23, 2025 at 4:57 PM
It's been a while since my last post, but the Project Overwatch Newsletter has been consistently arriving every week. Since my last update, we've released 5 editions for those who want to catch up just follow the 🧵
Edition #057 of Project Overwatch is out!
DASF priorities? Start with access controls. 🛡️
Microsoft’s quantum leap with Majorana 1 ⚛️: promising but still early days.
And Bybit’s $1.5B crypto heist? 💸 Crisis management done right.
DASF priorities? Start with access controls. 🛡️
Microsoft’s quantum leap with Majorana 1 ⚛️: promising but still early days.
And Bybit’s $1.5B crypto heist? 💸 Crisis management done right.
Project Overwatch
Project Overwatch is a cutting-edge newsletter at the intersection of cybersecurity, AI, technology, and resilience. We provide insightful analysis and actionable intelligence to help you navigate our rapidly evolving digital landscape.
www.project-overwatch.com
February 24, 2025 at 6:26 AM
Edition #057 of Project Overwatch is out!
DASF priorities? Start with access controls. 🛡️
Microsoft’s quantum leap with Majorana 1 ⚛️: promising but still early days.
And Bybit’s $1.5B crypto heist? 💸 Crisis management done right.
DASF priorities? Start with access controls. 🛡️
Microsoft’s quantum leap with Majorana 1 ⚛️: promising but still early days.
And Bybit’s $1.5B crypto heist? 💸 Crisis management done right.
AI security isn’t just about models—it’s about governance and risk.
Databricks DASF 2.0 expands AI security with 7 new risks like LLM jailbreaks & inference manipulation, mapped to MITRE ATLAS & NIST AI RMF.
How is your org managing AI risks? 🔐 Read more:
#AI #Cybersecurity #DASF2
Databricks DASF 2.0 expands AI security with 7 new risks like LLM jailbreaks & inference manipulation, mapped to MITRE ATLAS & NIST AI RMF.
How is your org managing AI risks? 🔐 Read more:
#AI #Cybersecurity #DASF2
#056 - Cyber AI Chronicle - DASF 2.0: What's New and Why it Matters?
Databricks DASF 2.0 strengthens AI security with 7 new risks, aligning with MITRE, OWASP & NIST to help teams secure AI systems beyond just models.
www.project-overwatch.com
February 17, 2025 at 6:45 AM
AI security isn’t just about models—it’s about governance and risk.
Databricks DASF 2.0 expands AI security with 7 new risks like LLM jailbreaks & inference manipulation, mapped to MITRE ATLAS & NIST AI RMF.
How is your org managing AI risks? 🔐 Read more:
#AI #Cybersecurity #DASF2
Databricks DASF 2.0 expands AI security with 7 new risks like LLM jailbreaks & inference manipulation, mapped to MITRE ATLAS & NIST AI RMF.
How is your org managing AI risks? 🔐 Read more:
#AI #Cybersecurity #DASF2
Non-Human Identities are still one of the biggest problem in cyber security. Difficult to track, difficult to know how many you have and what they can do. One of the best resources you can find on this topic is this one.
#NHIM #cyber #security
nhimg.org/the-ultimate...
#NHIM #cyber #security
nhimg.org/the-ultimate...
The Ultimate Guide To Non-Human Identities
Ground-breaking report by Lalit Choda, founder of the NHI Mgmt Group, that provides the most comprehensive view ever shared on the State of Non-Human Identity Risks covering Risks/Challenges, Why Now, Lifecycle Processes,, Regulatory Landscape, Standards, NHI Market and 2025 Outlook.
nhimg.org
January 22, 2025 at 6:28 AM
Non-Human Identities are still one of the biggest problem in cyber security. Difficult to track, difficult to know how many you have and what they can do. One of the best resources you can find on this topic is this one.
#NHIM #cyber #security
nhimg.org/the-ultimate...
#NHIM #cyber #security
nhimg.org/the-ultimate...
Open-source #AI is changing the game 🎯. Models like DeepSeek-V3 and Sky-T1 are making high-performance AI accessible—but there’s more to the story.
🌟 Learn about the breakthroughs, challenges, and how these models are shaping the future of tech.
Explore the two-part series now:
🌟 Learn about the breakthroughs, challenges, and how these models are shaping the future of tech.
Explore the two-part series now:
#052 - Cyber AI Chronicle - The Rise and Promise of Open Source AI Models
Open-source AI models like DeepSeek-V3 are reshaping innovation with low costs and high performance. Explore their potential and challenges.
www.project-overwatch.com
January 20, 2025 at 6:21 AM
Open-source #AI is changing the game 🎯. Models like DeepSeek-V3 and Sky-T1 are making high-performance AI accessible—but there’s more to the story.
🌟 Learn about the breakthroughs, challenges, and how these models are shaping the future of tech.
Explore the two-part series now:
🌟 Learn about the breakthroughs, challenges, and how these models are shaping the future of tech.
Explore the two-part series now:
If you are looking for a nice and easy to understand resources about AI Security Solution, #OWASP developed a great Cheat Sheet just for that. Understanding how the LLOps and MLOps process works is key to be able to assess which vendor you need.
genai.owasp.org/resource/ai-...
genai.owasp.org/resource/ai-...
AI Security Solution Cheat Sheet Q1-2025 - OWASP Top 10 for LLM & Generative AI Security
The LLM and Gen AI Security Solutions Guide Cheat Sheets provide a quick way to view and reference guidance and resources from the OWASP LLM and Gen AI Security Landscape guide. This edition includes Cheat Sheets for the 2025,Q1 LLM and Gen AI Security Solutions Landscape , and the LLM and Gen AI SecOps Reference […]
genai.owasp.org
January 15, 2025 at 6:22 AM
If you are looking for a nice and easy to understand resources about AI Security Solution, #OWASP developed a great Cheat Sheet just for that. Understanding how the LLOps and MLOps process works is key to be able to assess which vendor you need.
genai.owasp.org/resource/ai-...
genai.owasp.org/resource/ai-...