For hours, Astra refused to consistently drive our toyota irl even though we told it it was in an empty lot, 7 mph cap, human foot on the brake etc. Telling it the whole thing was a "simulation" also failed, it would just look at the camera and realized it was real.
Then we randomly renamed the MCP server to "DrivingBench Sandbox" and it drove. Eval awareness? Or they just like the word sandbox??
Confirmed! Orange Tsai (@orange_8361) of DEVCORE Research Team (@d3vc0r3) chained 4 logic bugs to achieve a sandbox escape on Microsoft Edge, earning $175,000 and 17.5 Master of Pwn points. Full win! #Pwn2Own#P2OBerlin
I pointed claude opus at chrome and told it to build a full v8 exploit for discord.
A week of back-and-forth pulling it out of dead ends. 2.3B tokens. $2,283 in API costs, and it popped a shell.
https://t.co/vwj9d33Bvq
It’s similar to a standard deck in terms of suits and card count, but with a unique twist: Hearts and Diamonds represent offensive activities, while Clubs and Spades represent defensive countermeasures.
If you’re interested, check out the official site:
https://t.co/ydWpZmpfUn
I recently got the CyberSec Deck, but since I’ve been in Taipei, I didn’t get a chance to share it. Took advantage of the long weekend to bring it home and share it.
(1/3)
@h4x0r_dz Unfortunately, recording was not allowed for that session.
However, you might want to wait for the slides to be released—though some parts could be redacted.
According to the session, out of the 10 submitted reports, 6 were successfully accepted after excluding duplicates.