GitHub slashes bug bounty payouts by up to 59% as AI-generated reports flood security teams

GitHub slashes public bug bounty payouts as AI report flood buries its security team

Code shack also putting new limits on first-time researchers, and reserving the biggest rewards for a hand-picked group of proven hunters GitHub has decided that, if everyone with an AI chatbot can file a bug bounty report, it may as well stop paying them like seasoned security researchers. Starting July 27, the Microsoft-owned code forge is overhauling its bug bounty program with a two-tier system that cuts rewards for public submissions while dangling much fatter payouts to a new invite-only group of researchers with proven track records. At the same time, newcomers will find themselves capped on how many reports they can submit until they've demonstrated they can produce something worth reading. The Microsoft-owned biz says that the shake-up is a response to the flood of low-effort and AI-generated reports now accompanying many bug bounty programs. Rather than paying more people to file more reports, GitHub wants to spend more on researchers who've proved they can find the real thing. Cathering Cassell, product security engineer at GitHub. "These changes are about two things: reducing the noise so we can focus on the signal, and building a program that serious researchers find rewarding to participate in." For researchers sticking with the public program, a low-severity finding that previously earned between $500 and $1,000 will now bring in $250. Medium bugs top out at $2,000 instead of $5,000, high-severity flaws have been cut from as much as $20,000 to $5,000, and the maximum reward for a critical vulnerability falls from $30,000 to $10,000. The highest rewards are now reserved for GitHub's new invite-only VIP program. There, low-severity findings are worth $1,000, medium bugs $7,500, high-severity issues $20,000, and critical vulnerabilities at least $30,000. Entry isn't open to everyone. GitHub says invitations will be based on a proven history of valid reports, with researchers needing anything from one accepted critical vuln to seven accepted low-severity findings to qualify. GitHubb is also enabling HackerOne's "signal requirement," limiting how many reports new researchers can submit before they've established a history of legitimate findings. It says genuine newcomers will still have up to four opportunities to prove themselves, while reports already sitting in the backlog will continue to be assessed under the previous payout structure. The overhaul follows changes GitHub introduced earlier this year that tightened report-quality requirements and warned researchers against flooding the platform with AI-assisted submissions. GitHub is betting that fewer reports, from researchers with a proven track record, will do more for security than an ever-growing pile of AI-assisted submissions waiting for someone to read them. ®

GitHub Cuts Public Bug Bounty Payouts, Moves Top Rewards to VIP Tier

Beginning July 27, 2026, GitHub will cut public bug bounty payouts by at least half at every severity level. Critical findings will drop from $20,000-$30,000+ to a fixed $10,000, while its permanent invite-only VIP tier will pay $30,000 or more. Reports filed before that date, including those already in GitHub's growing triage queue, will retain the previous payout terms. GitHub said the changes are intended to reduce noise while giving established researchers faster responses, higher rewards, and closer access to its security engineering team. "You don't earn more by submitting more," the company said. "You earn more by submitting better." The public program is moving from flexible ranges to fixed payments: * Low: $250, down from $617-$2,000 * Medium: $2,000, down from $4,000-$10,000 * High: $5,000, down from $10,000-$20,000 * Critical: $10,000, down from $20,000-$30,000+ The Hacker News calculated that the new public rates are 50% lower for medium, high, and critical findings and about 59% lower for low-severity reports when measured against the bottom of GitHub's previous ranges. GitHub said fixed payments should remove uncertainty and triage overhead, though it may still award discretionary bonuses for exceptional work. The VIP schedule sets payments at $1,000 for low-severity findings, $7,500 for medium, $20,000 for high, and $30,000 or more for critical vulnerabilities. Researchers can qualify for the private program by reporting at least one critical, two high, four medium, or seven low-severity vulnerabilities. The announcement does not specify a time window for meeting those thresholds or say whether qualification guarantees an invitation. GitHub said fuller criteria will appear on its public HackerOne program page. GitHub has not disclosed the HackerOne Signal threshold it will enforce. The company says researchers below it will receive up to four initial submissions. Separately, HackerOne's general rules give new researchers four trial reports per program within a rolling 30-day window. When Your Own AI Finds the Bug First GitHub's report controls arrive as AI makes candidate findings cheaper to generate and code review cheaper to repeat. More researchers can produce potential findings, while internal teams can scan code, validate issues, and feed fixes into release and commit pipelines before an external report arrives. A day before GitHub's announcement, Google introduced Gemini 3.5 Flash Cyber, a lightweight model fine-tuned to find, validate, and patch software vulnerabilities. Google said the model will initially be available exclusively to governments and trusted partners through CodeMender, its code-security agent, as part of a limited pilot. Google said the model can be invoked repeatedly to examine more code paths without using a larger frontier model for every attempt. The company positions it for frequent repository scans, time-sensitive launch reviews, and commit-scanning pipelines. In Google-run tests, Gemini 3.5 Flash Cyber found 55 unique confirmed V8 issues, compared with 47 for mainline Gemini 3.5 Flash and 36 for Claude Opus 4.6. Google separately said its Cloud Vulnerability Research team used the model to find remote code execution flaws in public APIs and a memory-corruption flaw in a sensitive production service within two hours. The model then generated what Google described as a 100%-reliable RCE exploit that bypassed ASLR and W^X. The benchmark figures and production exploit result are Google-reported and have not been independently verified. An internal security team can give an agent repository context, a project-specific threat model, and a validation environment tailored to the running system. Systems such as OpenAI's Codex Security can then test findings, generate working proofs of concept, and propose fixes that account for system intent and surrounding behavior. The work can happen during development and on every relevant commit, rather than waiting for a scheduled assessment or an external report. AI does not replace a penetration test, but source-code review, test generation, and first-pass validation are becoming easier to automate. Human testers retain more value where the work requires chaining weaknesses across trust boundaries, recognizing business-logic failures, modeling realistic attack paths, and proving material impact. Curl maintainer Daniel Stenberg ended the project's cash bug bounty at the end of January 2026 after its confirmed-vulnerability rate fell below 5% amid an increase in AI-generated junk reports. By April, after curl had ended cash rewards and returned to HackerOne, reports were arriving at about twice the 2025 rate and 15-16% were confirmed as vulnerabilities. Stenberg said almost every report appeared AI-assisted and most were now high quality. Taken together, GitHub's report controls, Google's repeated model calls, and curl's rising submission volume point to the same shift. AI can flood maintainers with junk, but it can also make capable researchers faster and let internal teams examine more code, more often. A plausible-looking candidate finding is becoming abundant. Triage, exploit proof, product context, disclosure, and remediation remain constrained. A reliable exploit, a product-specific attack chain, or a finding that crosses a boundary the vendor misunderstood remains scarce. Signal requirements and lower public rewards may suppress automated noise, but they can also make entry harder for capable researchers without an established HackerOne history. For a new HackerOne researcher, a four-report program limit leaves little room for mistakes, unfamiliarity with GitHub's security model, or a legitimate finding that is initially scored below expectations. The invite-only structure also concentrates GitHub's closest researcher relationships among people who have already succeeded inside the program. That may improve speed and report quality. It may also narrow the range of people examining the platform, one of the main advantages of a public bounty program. The restructuring follows a May 2026 policy change that demanded working proofs of concept, demonstrated impact, validation before submission, and closer attention to GitHub's scope and ineligible findings. GitHub said it welcomes AI-assisted security research and already uses AI across its internal security programs. Researchers remain responsible for reproducing and verifying anything their tools produce. "The tools don't matter," GitHub said. "The quality of the work does." As of July 22, a review by The Hacker News found that GitHub's rewards page still listed $20,000-$30,000+ for critical reports, while its FAQ retained the previous VIP eligibility test of at least $20,000 earned and two reports submitted during the preceding two years. The FAQ also said meeting those criteria did not guarantee an invitation and that GitHub reviewed candidates quarterly.