Max Andreacchi

This One Weird Trick: Multi-Prompt LLM Jailbreaks (Safeguards Hate It!) - SpecterOps Using multiple prompts within the context of a conversation with an LLM can lead to safeguard bypasses. Learn about safeguards evaluations at scale.

Spoiler alert: Your AI safety measures might have a blind spot. 👀

When attackers use conversation context to bypass LLM safeguards, single-prompt evals just don't cut it anymore.

Dive into @atomicchonk.bsky.social's latest blog on multi-prompt attack detection. ghst.ly/47qJhzn

Potato exploits have been a cornerstone of local priv esc on Windows for years, but how & why do the inner starchy workings of the potatoes function?

Join @atomicchonk.bsky.social next week to understand Windows access tokens & their use in the Windows environment. ghst.ly/june-web-bsky

Sorry, Italian four cheese and cheddar jack supreme. My mind was still recovering from the abomination

Idk if it’s the Cheez It crust itself for me, or if it’s the fact that someone thought a Cheez It crust and an Italian Supreme flavor were a good mix

Tokenization Confusion - SpecterOps Meta's Prompt Guard 2 aims to prevent prompt injection. This post looks at how much knowledge of ML we need to be effective at testing these LLM WAFs.

New blog post is up! Stepping out of my comfort zone (be kind), looking at Meta's Prompt Guard 2 model, how to misclassify prompts using the Unigram tokenizer and hopefully demonstrate why we should invest time looking beyond the API at how LLMs function. specterops.io/blog/2025/06...

You’re a real one, thank you 🙏🏼

It's potato harvest season! 🥔

Join our upcoming webinar w/ @atomicchonk.bsky.social as he breaks down the starchy workings of potato exploits — from Windows access tokens to technical walkthroughs of Rotten, Juicy, and Rogue potatoes.

Register at ghst.ly/june-web-bsky

Just wrapped up a blog post on understanding BadSuccesor from a DACL abuse aspect and mitigating it from a DACL abuse perspective.

Also added some PowerShell on my GitHub to create and remove the mitigations.

Understanding & Mitigating BadSuccessor - SpecterOps Understanding the impact of the BadSuccessor AD attack primitive and mitigating the abuse via targeted Deny ACEs on Organizational Units.

BadSuccessor is a new AD attack primitive that abuses dMSAs, allowing an attacker who can modify or create a dMSA to escalate privileges and take over the forest.

Check out @jimsycurity.adminsdholder.com's latest blog post to understand how you can mitigate risk. ghst.ly/4kXTLd9

If you haven’t read the BadSuccessor blog post, woo boy: www.akamai.com/blog/securit...

Ok I’ve been enlightened further; they don’t collide if you tweak the command just slightly :pain:

a rainbow and a star with the words " make you know " on it ALT: a rainbow and a star with the words " make you know " on it

I’m convinced most learning happens when you’re doing what I call “smacking into something;” failing repeatedly, figuring out why it failed, and proceeding to the next step where you rinse and repeat until you achieve your ultimate objective. TIL: docker and podman dependencies collide.

Congratulations!

Black Hat USA 2025

Beyond hyped to be presenting with @anam0x.bsky.social and the rest of my team at Arsenal at BHUSA 2025! app.ingo.me/q/0x9xn

Squeezing Cobalt Strike Threat Intelligence from Shodan One of my favorite Twitter accounts from the last several years was @cobaltstrikebot, mainly because it was an awesome source of threat intelligence for Cobalt Strike beacons in the wild. The account ...

Do you miss "@cobaltstrikebot"? If so, here's a blog post showing how you can pull Cobalt Strike SpawnTo and watermark info with @shodanhq.bsky.social and some PowerShell: forensicitguy.github.io/squeezing-co...

It was an absolute pleasure to speak at @cackalackycon.bsky.social today and share my love of potatoes. Thank you to @specterops.io for fueling me to always go a layer deeper in learning and motivating me to chase my passions 🥔

What do potatoes have to do with privilege escalation on Windows? Come find out at Max Andreacchi’s session, “Tater Tokens: Introduction to Windows Access Tokens and Their Role in PrivEsc” on May 18th!

Why do potato exploits work & how can we stop them?

Join @atomicchonk.bsky.social at @cackalackycon.bsky.social this weekend for a walkthrough of Windows access token manipulation and get the answer. ghst.ly/4jzjlnI

Immaculate Rick Roll placement

Great work getting it done!

Always enjoy the views in Seattle! Spent excellent quality time with teammates and received amazing training. Now for a weekend of running and resting back home before new travels next week ✈️

Don't let threat actors mash your Windows security! @atomicchonk.bsky.social’s @cackalackycon.bsky.social talk breaks down potato exploits from token mechanics to defensive implementations.

Learn more ➡️ ghst.ly/4jzjlnI

Fulltime Win Graphic: NC Courage 3-2 KC Current.

NC BABYYY 💙

I’ve had to beat stubbornness out of my training. When I ran 10+ yrs ago I’d just hard-head my way forward but wind up injured. This time I’m focused on the goals ahead; nagging aches mean a rest day tomorrow so I can live to run later this week and keep this train moving to Oct (and beyond)

Understanding Windows access tokens could be your best defense. At @cackalackycon.bsky.social, @atomicchonk.bsky.social will be peeling back the layers on potato exploits that threat actors use for privilege escalation.

Check out the schedule to learn more ➡️ ghst.ly/4jzjlnI

Just pushed a new versions for #AADInternals and AADInternals-Endpoint modules! Some bug fixes plus support for:
1️⃣ Microsoft Authentication Library (MSAL)
2️⃣ Token Protection
3️⃣ Continuous Access Evaluation (CAE)

DPRK IT Workers in Open Source and Freelance Platforms A cluster of actors discovered in onlyDust.com freelancer platform and beyond

The Ketman Project has published a list of names and GitHub profiles they believe may be North Korean rogue IT workers posing as open-source developers and freelancers, and seeking employment at Western software companies

www.ketman.org/dprk-it-work...

Initial stab at using Chris Hayuk's mcp-cli tool to pair roadrecon_mcp_server with a locally-hosted model (in this case, mistral-small3.1). Note that running this query took over 300s so YMMV depending on several factors. Will update GH this week with instructions on replicating this.