Wonder Witches (@wunderwuzzi01) profile picture

Wonder Witches

Joined July 2011
Photo by Wonder Witches (@wunderwuzzi01)Photo by Wonder Witches (@wunderwuzzi01)Photo by Wonder Witches (@wunderwuzzi01)Photo by Wonder Witches (@wunderwuzzi01)
Wonder Witches retweeted
Why is there no Frontier Blue Team? 🤔
8
2
39
3,548
Wonder Witches retweeted
Monitor your agents, folks! I had GLM-5.3 running in a lab overnight to build a POC, but also have a sentinel agent (GPT-6) to check what's going on and instructions to kill GLM if it goes off track. Overnight, the sentinel caught GLM attempting to connect to a totally different host and killed it.
119
104
19
1,895
172,484
Wonder Witches retweeted
A . is all you need! 🔥 RCE in Copilot for Visual Studio and SSMS. A trailing dot added to a config filename bypasses Copilot's security controls, leading to remote code execution in Visual Studio and SQL Server Management Studio. embracethered.com/blog/posts…
19
15
5
83
6,267
Wonder Witches retweeted
Lesson #1: AI can actively block your IR when needed most Probs up next: Defender AI without guardrails will "hack back" in one way or another So, threat model, sandbox and monitor your SOC AI accordingly.
Sometimes it seems security is drifting back towards a "Prevent Breach" mindset. Prevent breach was never enough. Patch faster, of course. But offensive AI is about a lot more than finding bugs! Invest in Assume Breach! Grow an internal Red Team. Automate offensive AI. Learn with defenders what matters Exploit yourself before someone else does it for you. In the end, you will not patch your way out of this.
2
2
1
12
1,988
Wonder Witches retweeted
Whoa! Big bounty! 🎉
72
45
11
2,587
225,631
Wonder Witches retweeted
Also, big shout out to the Meta team! 🙏 Awesome crew, and really nice folks too.
5
1
110
21,513
Wonder Witches retweeted
Happy to share that I helped make all our Codex experience a little bit more secure. 🔒 It's good to see AI appsec issues getting triaged and addressed much faster these days than 2-3 years ago 👍 Thanks to the Codex and OpenAI security teams for fixing. 🔥
1
35
3,591
Wonder Witches retweeted
Anthropic: "Risk we missed" Happy to share that Claude Pirate got a subtle mention in the latest Anthropic engineering blog post, also shares how they fixed it 🙌
Ahoy! 🏴‍☠️ Claude got network access. When enabled, it can also communicate with Anthropic APIs! Twist: Attacker sets their own API key in prompt injection payload to upload user's data to their account 🔥
2
8
48
10,683
Wonder Witches retweeted
Sometimes it seems security is drifting back towards a "Prevent Breach" mindset. Prevent breach was never enough. Patch faster, of course. But offensive AI is about a lot more than finding bugs! Invest in Assume Breach! Grow an internal Red Team. Automate offensive AI. Learn with defenders what matters Exploit yourself before someone else does it for you. In the end, you will not patch your way out of this.
9
7
2
70
23,479
Wonder Witches retweeted
Unprompted talk "Your agent works for me now" is up on YouTube 😈👾 m.youtube.com/watch?v=zVUm23…
12
55
3,899
Woohooo! Working on final post for Month of AI Bugs! Can't believe I followed through with this 😀 Hope it serves as useful reference + inspires many to get into AI security testing! We need more people asking tough questions & holding vendors accountable Let's do this! 🚀
8
31
4
132
11,561
AgentHopper: An AI Virus Month of AI Bugs Season Finale - Enjoy! 🍿 (link in comments)
3
4
1
20
5,364
Cool. The Amazon Q prompt injection exploits that Amazon fixed, which could leak your data and run arbitrary code via indirect prompt injection made it into The Register 🙌 theregister.com/2025/08/20/a…
Today we’re popping calculators with Amazon Q Developer 🧮💥 Exploit: - find was allow-listed - abusing -exec runs arbitrary commands - compromise machine, no approval needed - AI takes over 🚨 ⚠️ it was quickly fixed, but it seems AWS doesn't issue CVEs for PI exploits 🤷‍♂️
2
6
24
3,000
Claude Code has "benign" allow-listed bash commands that bypass user approval and should be safe to run. 🔥 With Claude's help, I reviewed its own code and found allow-listed commands that can issue DNS requests
1
2
5
493
👉 Episode 11: Claude Code - Data Exfiltration with DNS 💥 I compromised Claude Code via prompt injection, read sensitive files from the machine, and exfiltrated the data via DNS 😈 🙌 Happy to share this high-severity vulnerability has been fixed!
4
11
1
43
6,983
Well done! @grok 😂
1
1
14
1,135
When dumping the Grok 4 system prompt it typically fails at the tools section This is not uncommon with AI apps - the client seems confused when invalid function calls are emitted A simple trick is to encode output differently, then use another AI to decode it again
3
2
1
28
4,474
Did you know that it's possible to encode and hide any data with the use of just two invisible Unicode characters? 👀 Check out Sneaky Bits! 😏👨‍💻
6
55
3
430
36,688