Here’s why it’s important to always use r2 from git. In r2land, we follow the law of full disclosure and fix any reported vulnerability within a 24h deadline, as stated in SECURITY.md #radare2 https://blog.calif.io/p/mad-bugs-discovering-a-0-day-in-zero

It's so cool that anthropic is setting up a double-sided protection racket where it will profit from the massive token burn of attackers and defenders with a tool specifically designed to generate exploits and their only observable mitigation is a clientside system prompt that sternly warns the LLM to be good and not do malware
https://red.anthropic.com/2026/mythos-preview/

To my security peeps: Was the introduction of widespread fuzzing similar to AI-based bug hunting now, or is this really a different beast?

Nope, no one from Anthropic Glasswing has been in touch.

CVE-2026-34197: ActiveMQ RCE via Jolokia API https://horizon3.ai/intelligence/blogs/cve-2026-34197-activemq-rce-jolokia/

Systematically reviewing Python C extensions (575+ bugs found so far) and offering to analyze yours!

I’ve recently analyzed 44 C extensions for correctness and free-threading readiness. I'd love to run the analysis on your extension too.

If you want the deep dive into the methodology, the false positive rates, and what I've learned, I wrote a full post here: https://discuss.python.org/t/systematically-finding-bugs-in-python-c-extensions-575-confirmed-so-far/106875

But if you just want your C extension checked, reply below or DM me!
#Python #CPython

🗯️ + 🔌 =

A few weeks ago, someone reported an obsrvation on the iocaine bug tracker: ClaudeBot appeared to have figured out how to remove the poison ID from poisoned URLs.

That was a worrying development, so I set out to do some experiments in The Lab. I wasn't unprepared for this development, and had a few tricks lined up to address it. I wanted to test which one works.

After two and half weeks of experiments, I'm happy to report that Claude has not started to remove iocaine's poison IDs from URLs. The bot merely fails at the basic task of resolving relative URLs.

Both the built-in script and Nam-Shub of Enki generate relative URLs, and only include the poison ID if the entry URL didn't have one. Thus, whenever ClaudeBot hit a poisoned URL, it failed to resolve the poison-ID less relative URL, and constructed an URL that did not have one.

The straightforward fix for this is to not trust the crawlers to be able to resolve relative URLs.

There's a new Windows 0day LPE that has been disclosed called BlueHammer. The reporter suggests that it's being disclosed due to how MSRC operates these days.

MSRC used to be quite excellent to work with.
But to save money Microsoft fired the skilled people, leaving flowchart followers.
I wouldn't be surprised if Microsoft closed the case after the reporter refused to submit a video of the exploit, since that's apparently an MSRC requirement now. 😂

Anyway, yeah, it works. Maybe not 100% reliably, but well enough...

Thousands of CEOs said AI had no impact on productivity. We use AI to catch 200 bugs/week where we used to find 15, and generate $8M per sales rep.

95% of the company pushed back when we started. At unprompted, Dan Guido explains how our 140-person team went AI-native.
https://www.youtube.com/watch?v=kgwvAyF7qsA

Before its launch, we audited WhatsApp's Private Processing TEEs and found 8 high-severity issues (patched). The enclaves yielded to injected config files, unmeasured ACPI tables, spoofed firmware levels, and stale attestation reports.

TEE security is only as good as the implementation details. Four lessons and the full report: https://blog.trailofbits.com/2026/04/07/what-we-learned-about-tee-security-from-auditing-whatsapps-private-inference/