Wrote up some notes on PCIe DMA cheats — how the protocol stack works, how IOMMU translation and bypass techniques interact, what FPGA hardware the attack surface looks like, and how detection stacks up from config-space fingerprinting through TPM attestation.
https://t.co/gjQn6XWwFk
Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946.
For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids.
An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better.
This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.
openai․com was once the personal homepage of a guy named glenn (2001), and tiktok․com was the quaint shared homepage of a couple who went from dating to married with a baby (~1998-2001). THREAD!!!
New research explains why LLMs hallucinate, through a connection between supervised and self-supervised learning. We also describe a key obstacle that can be removed to reduce them. 🧵https://t.co/6Lb6xlg0SZ
While playing @defcon CTF Finals with @shellphish I managed to solve the ICO challenge using LLMs (GPT5 + Cursor) and almost no human intervention. You can read how I did it here! https://t.co/EcqYZdyIfV
Pwn2Own Berlin 2025 comes to a close. We awarded $1,078,750 for 28 unique 0-days. Congrats to @starlabs_sg for winning Master of Pwn with $320,000. Thanks to @offensive_con for hosting, and thanks to all who participated. Can't wait to see you next year! #Pwn2Own#P2OBerlin