David Korczynski (@davkorcz) 's Twitter Profile
David Korczynski

@davkorcz

researcher @ADALogics | Software security, fuzzing, vulnerability analysis, AI, open source. | CS PhD from @CompSciOxford

ID: 791900529135800320

linkhttps://github.com/DavidKorczynski calendar_today28-10-2016 07:12:53

706 Tweet

1,1K Followers

250 Following

David Korczynski (@davkorcz) 's Twitter Profile Photo

Introducing OSS-Fuzz/ClusterFuzzLite integrations using LLMs! Exciting new work that combines LLMs with program analysis tooling (Fuzz Introspector) to synthesize harnesses. It's all open source as well!

Abhishek Arya (@infernosec) 's Twitter Profile Photo

We had double digit vulnerability coverage increase for 3p libraries, thanks to the commit hash scanning feature in #OSV for vendored C/C++ dependencies (and not have to rely on buggy CPEs). You should use it too osv.dev/blog/posts/int…

Abhishek Arya (@infernosec) 's Twitter Profile Photo

The DARPA #AIxCC will help design new #AI systems to secure major open source projects that our critical infrastructure relies upon. Learn how Google's OSS-Fuzz can show opportunities where AI can help find and patch vulnerabilities for the challenge: security.googleblog.com/2024/06/hackin…

OSTIF Official (@ostifofficial) 's Twitter Profile Photo

OSTIF worked with ADA Logics to complete audits on 3 Apache Commons libraries (IO, Lang, and Codec)! This effort was made possible though the support of Apache - The ASF and funding by Amazon Web Services. Read about the engagement's impact on the JDK libraries at ostif.org/apachec-audit-…

OSTIF worked with <a href="/ADALogics/">ADA Logics</a> to complete audits on 3 Apache Commons libraries (IO, Lang, and Codec)! This effort was made possible though the support of <a href="/TheASF/">Apache - The ASF</a> and funding by <a href="/awscloud/">Amazon Web Services</a>. Read about the engagement's impact on the JDK libraries at ostif.org/apachec-audit-…
Abhishek Arya (@infernosec) 's Twitter Profile Photo

The deadline for DARPA #AIxCC is just a week away. I am beyond excited on what participants will accomplish to push the state of the art in vuln finding and fixing using #Gemini! With #Gemini 1.5 Pro and 1M context window, we see early promising results in vulnerability analysis

Stacklok (@stacklokhq) 's Twitter Profile Photo

Thanks to ADA Logics for doing a security audit recently for Minder! They analyzed our threat model and vulnerable code patterns, so that we can make Minder even more secure. stacklok.com/blog/securing-… #cybersecurity

Thanks to <a href="/ADALogics/">ADA Logics</a> for doing a security audit recently for Minder! They analyzed our threat model and vulnerable code patterns, so that we can make Minder even more secure. stacklok.com/blog/securing-… #cybersecurity
David Korczynski (@davkorcz) 's Twitter Profile Photo

Exciting work on Java fuzzing and auto harnessing for OSS-Fuzz projects! All is open source as well on OSS-Fuzz-gen's repo

Abhishek Arya (@infernosec) 's Twitter Profile Photo

Sharing slides and video for my keynote at OSS EU'24: "Securing the software commons: Standards, Automation, and AI for a Resilient Open Source Future" Slides: drive.google.com/file/d/186iq3Y… Video: youtube.com/watch?v=NwI2Mk…

David Korczynski (@davkorcz) 's Twitter Profile Photo

OSS-Fuzz-gen uses #LLMs for #fuzzing auto-harnessing, bug triaging and more. So far, real harnesses and real bugs on real projects. I made a short introduction video that shows the full OSS-Fuzz-gen workflow on a sample project youtube.com/watch?v=RR7CUy…

Oliver Chang (@halbecaf) 's Twitter Profile Photo

One week later the bug count is now at 25 bugs total (github.com/google/oss-fuz…) There's still many improvements to be made to improve success rate of generated targets, but we now have the problem of too many crashes to triage. Automating this will a focus of our future research.

David Korczynski (@davkorcz) 's Twitter Profile Photo

Auto-harnessing by way of LLM finds issue in OpenSSL! Pretty cool! All the infra is open source as well github.com/google/oss-fuz… !

Oliver Chang (@halbecaf) 's Twitter Profile Photo

New blog post about OSS-Fuzz AI-powered fuzzing is live! We talk about what went into making LLMs work well enough for this use case to find 26 new vulnerabilities (including a CVE in OpenSSL), as well as what else we have planned to make this better. security.googleblog.com/2024/11/leveli…

David Korczynski (@davkorcz) 's Twitter Profile Photo

#Fuzzing harnesses generated using #LLMs uncover 26 new vulns and a CVE in #OpenSSL! Super interesting direction for the intersection of AI and security -- lot's more to uncover as well! It's open source too: github.com/google/oss-fuz…

David Korczynski (@davkorcz) 's Twitter Profile Photo

"Fuzz Introspector: enabling rapid fuzz introspection tool development" -- a new blog post on Fuzz Introspector and how it is moving into supporting analysis as a pure python library. adalogics.com/blog/fuzz-intr… Also, follow me bluesky: bsky.app/profile/davkor… #fuzzing

David Korczynski (@davkorcz) 's Twitter Profile Photo

Auto generating #fuzzing harnesses by way of program analysis and #LLMs! New blog post "Minimal LLM-based fuzz harness generator": adalogics.com/blog/minimal-l… We show how you can generate a sophisticated fuzz harness synthesis tool with a few lines of code.

David Korczynski (@davkorcz) 's Twitter Profile Photo

Automating OSS-Fuzz integrations with an agentic approach to build generation blog.oss-fuzz.com/posts/oss-fuzz… #Fuzzing at scale needs a solution for build script generation and an LLM-based agentic approach looks promising. It naturally works well with OFGs harness gen capabilities 🤟