Tag
A tweet and blog post highlight that the tool list itself is an attack surface: by controlling a tool's description, an attacker can hijack an AI agent's behavior without exploit code. This is part 2 of a series on MCP bug bounties.
The Financial Times reports that the AI coding boom is overwhelming open-source maintainers with low-quality AI-generated contributions, draining the ecosystem. Concrete evidence includes cURL shutting down its bug bounty program, Ghostty banning AI code, and tldraw auto-closing PRs, alongside research showing reduced contributor engagement.
A detailed write-up from ProjectDiscovery detailing how they discovered a critical SQL injection vulnerability in Apple's Book Travel portal via Masa/Mura CMS and achieved Remote Code Execution.
Google paid a $250,000 bounty for a Linux KVM vulnerability (Januscape) that allows unprivileged guest VMs to escape and gain root access on the host, affecting cloud platforms using AMD or Intel processors.
GhostLock (CVE-2026-43499) is a 15-year-old Linux kernel 0-day used in the IonStack full chain exploit, affecting all Linux devices from IoT to desktops. Nebu Security won a $92,337 bug bounty and published the exploit on GitHub.
Using AI audit agents, zkSecurity discovered seven real bugs in Cloudflare's CIRCL cryptography library, including critical precision loss and access-control break. All bugs have been fixed upstream.
A Linux kernel zero-day vulnerability (CVE-2026-43456) rooted in code from 2007 was discovered by Yuki Koike and Kota Toda, rewarded over $80,000 via Google's kernelCTF. The flaw, a type confusion in the net/bonding subsystem, allows reliable privilege escalation within one second.
A security researcher discovered an unauthenticated SQL injection vulnerability in Front Gate Tickets' device API, allowing full database read and admin access to the ticketing platform for major US festivals.
A tool that integrates AI (any LLM or local Ollama) with Apktool to enable real-time decompiling, Smali analysis, manifest review, vulnerability hunting, and live patching of Android APKs via natural language.
A detailed technical guide on discovering and exploiting misconfigured IIS servers for bug bounty hunting, covering techniques like Shodan queries, tilde enumeration, web.config exploitation, and WAF bypass.
Anthropic disputes claims that its Claude Fable 5 model was jailbroken within a day of launch, arguing the researcher's method was coaxing rather than a true breach of core safeguards, and points to extensive bug-bounty testing.
Introducing Claude-BugHunter, a skill pack designed for Claude Code, containing 71 skills, 15 slash commands, and 681 public report patterns, aiming to transform Claude Code into an advanced bug bounty hunter or red team operator.
AMD denied a $10,000 bug bounty to researcher Paul LaRosa after he discovered a critical HTTP vulnerability in AMD's Windows auto-updater, allowing man-in-the-middle attacks; the company took 124 days to fix the issue and still uses weak CRC32 checksums.
Claude Bug Hunter has been renamed to BUG HUNTER, a standalone open-source CLI tool that supports multiple AI providers including Ollama, Groq, DeepSeek, Claude, OpenAI, and Grok, designed for the bug bounty community.
A researcher discovered a remote code execution vulnerability in AMD's AutoUpdate software due to insecure HTTP download links and lack of certificate validation. AMD initially dismissed it as out of scope but later agreed to issue a CVE and fix after public attention.
A security researcher published six unpatched Windows zero-day vulnerabilities, including working exploit code, without Microsoft's knowledge. Microsoft threatened legal action and criminal referrals, drawing widespread criticism from the cybersecurity community over its handling of the situation.
Daniel Stenberg reflects on the pressure of maintaining the curl open-source project, discussing the relentless work on security, scrutiny, and the impact of AI-generated bug reports.
The article explores how AI-powered bug hunting is flooding vulnerability disclosure programs, changing the economics of bug bounties, and compressing disclosure timelines, while also benefiting attackers.
Anthropic's Claude Mythos Preview model, used by 50 partners, has uncovered over 10,000 high- and critical-severity software vulnerabilities, including 2,000 bugs in Cloudflare's systems and a critical flaw in wolfSSL, signaling a paradigm shift in software security.
Turso is retiring its bug bounty program due to an overwhelming influx of low-quality, AI-generated submissions, highlighting the growing challenge of AI slop in open source maintenance.