“AI has made discovering bugs low cost, however judging them continues to be costly,” Leap’s Ishan Bhatt wrote in an Oct. 16 assessment.
Engineers nonetheless needed to reproduce reported issues and test whether or not the proposed assaults may work. Some submissions flagged obvious weaknesses whereas lacking protections elsewhere within the software program.
The backlog peaked at 299 open experiences. As critiques slowed, different researchers rediscovered bugs that had already been submitted however remained unfixed or undisclosed. About 80% of duplicates arrived greater than three days after the unique report.
The competitors had already charged $100 per submission and restricted researchers to 1 report a day. Leap says charging discourages researchers from submitting massive numbers of speculative findings and leaving its engineers to validate them.
Its new platform additionally limits every principal report part to three,000 characters.
The competitors uncovered severe vulnerabilities, together with a scheduling error that would make Firedancer execute transactions within the flawed order and disagree with different Solana validators concerning the ensuing community state.
In the meantime, Leap is rising rewards alongside the submission price. Eligible findings obtain twice the usual payout by way of Nov. 20, or thrice for bugs particular to Alpenglow, Solana’s new voting system. Important Alpenglow findings can qualify for as much as $300,000.