[
	{
		"id": "bsides26-bugbounty-1000",
		"title": "The AI Submission Surge: Vulnerability Disclosure Programs in the Age of AI",
		"description": "Generative AI is changing vulnerability disclosure programs on both sides of the submission queue. It helps experienced researchers understand targets faster, structure proofs of concept, and communicate findings more clearly. It also makes polished but unverified vulnerability claims inexpensive to produce at scale.\n\nThis defender-side field report examines how that shift affects vulnerability intake, evidence validation, reviewer workload, and program design. Using sanitized and aggregated patterns from large-scale vulnerability disclosure operations, we will show why credible-looking reports can no longer be treated as credible findings, where AI-assisted research provides genuine value, and how disclosure programs can filter low-signal claims without slowing down strong researchers.",
		"time": {
			"start": "2026-11-11T10:00:00",
			"end": "2026-11-11T10:45:00"
		},
		"speakers": [
			{
				"id": "justin-roy",
				"name": "Justin Roy"
			}
		],
		"hall": {
			"id": "Bug Bounty Village - BSides",
			"name": "Bug Bounty Village - BSides"
		},
		"tags": [],
		"track": "Bug Bounty Village - BSides",
		"embeddings": {
			"embedTextHash": "-1989811638",
			"generatorVersion": "-1254175146"
		},
		"otherData": {
			"type": "talk",
			"year": 2026
		},
		"contentHash": "tl0fxx-1ga7cfz"
	},
	{
		"id": "bsides26-bugbounty-1045",
		"title": "Open Redirect: From “Low Severity” to Real-World Impact",
		"description": "Open Redirects are often considered low-severity vulnerabilities. But how much can you actually do with one?\n\nWe’ll start with a simple phishing example, showing how a trusted domain can be used to redirect victims to an attacker-controlled login page. Then we’ll dive into a real bug bounty case where an Open Redirect was embedded into a legitimate email flow.\n\nThe main focus will be on the remediation: the target deployed a fix, I was asked to verify it, and found multiple ways to bypass it. We’ll examine the URL parsing mistakes behind the bypasses, using real screenshots and recordings from the report.\n\nFinally, we’ll look at another real-world case that demonstrates the same vulnerability class.",
		"time": {
			"start": "2026-11-11T10:45:00",
			"end": "2026-11-11T11:30:00"
		},
		"speakers": [
			{
				"id": "avishai-gonen",
				"name": "Avishai Gonen"
			}
		],
		"hall": {
			"id": "Bug Bounty Village - BSides",
			"name": "Bug Bounty Village - BSides"
		},
		"tags": [],
		"track": "Bug Bounty Village - BSides",
		"embeddings": {
			"embedTextHash": "1124048955",
			"generatorVersion": "-1254175146"
		},
		"otherData": {
			"type": "talk",
			"year": 2026
		},
		"contentHash": "1xsyq6y-un7pjs"
	},
	{
		"id": "bsides26-bugbounty-1140",
		"title": "Ahead of the Curve: Scaling Bug Bounty Program in the Age of AI",
		"description": "How is AI changing vulnerability research and how should a bug bounty program respond? This talk draws on 15 years of Meta's Bug Bounty program to cover the shift to agentic AI triage, what it means for researchers and why hacking Meta pays off.",
		"time": {
			"start": "2026-11-11T11:40:00",
			"end": "2026-11-11T12:00:00"
		},
		"speakers": [
			{
				"id": "roman-lifshits",
				"name": "Roman Lifshits"
			}
		],
		"hall": {
			"id": "Bug Bounty Village - BSides",
			"name": "Bug Bounty Village - BSides"
		},
		"tags": [],
		"track": "Bug Bounty Village - BSides",
		"embeddings": {
			"embedTextHash": "1080376549",
			"generatorVersion": "-1254175146"
		},
		"otherData": {
			"type": "talk",
			"year": 2026
		},
		"contentHash": "cf1ct7-sy8ned"
	},
	{
		"id": "bsides26-bugbounty-1200",
		"title": "Beyond the Horizon: Breaking Azure's DBaaS from the Inside",
		"description": "What happens when cloud providers start embedding AI directly inside your database? New features mean a new attack surface - and nobody's really looking at it yet.\n\nAzure recently shipped an AI extension inside their managed PostgreSQL service, giving every database admin access to built-in AI functions powered by Azure OpenAI. These functions run under a hidden platform superuser. They accept user input. And they trust it.\n\nDuring authorized research under Microsoft's bug bounty program, I found a chain of vulnerabilities in this AI extension that took me from a regular database admin to a full reverse shell on the underlying container - earning CVE-2026-85878 with a 9.9 Critical severity score. Four security boundaries crossed.\n\nThis talk walks through the entire chain step by step: the discovery, the dead ends, the PostgreSQL internals trick that made it all work, and what I found after getting the shell. I'll also share a general methodology for attacking any managed database service - patterns that apply whether you're hunting on Azure, AWS, or GCP.",
		"time": {
			"start": "2026-11-11T12:00:00",
			"end": "2026-11-11T12:25:00"
		},
		"speakers": [
			{
				"id": "uriel-yochpaz",
				"name": "Uriel Yochpaz"
			}
		],
		"hall": {
			"id": "Bug Bounty Village - BSides",
			"name": "Bug Bounty Village - BSides"
		},
		"tags": [],
		"track": "Bug Bounty Village - BSides",
		"embeddings": {
			"embedTextHash": "-953651232",
			"generatorVersion": "-1254175146"
		},
		"otherData": {
			"type": "talk",
			"year": 2026
		},
		"contentHash": "1ryziuf-ilxnfl"
	},
	{
		"id": "bsides26-bugbounty-1300",
		"title": "Lunch Break",
		"description": "",
		"time": {
			"start": "2026-11-11T13:00:00",
			"end": "2026-11-11T14:10:00"
		},
		"speakers": [],
		"hall": {
			"id": "Bug Bounty Village - BSides",
			"name": "Bug Bounty Village - BSides"
		},
		"tags": [],
		"track": "Bug Bounty Village - BSides",
		"type": "break",
		"embeddings": {
			"embedTextHash": "1928093769",
			"generatorVersion": "-1254175146",
			"facets": {
				"values": [],
				"phrasesHash": "0",
				"generatorVersion": "-1254175146"
			}
		},
		"otherData": {
			"type": "break",
			"year": 2026
		},
		"contentHash": "11e3pfc-1omshr6"
	},
	{
		"id": "bsides26-bugbounty-1410",
		"title": "Upscaling end-to-end bug claim response",
		"description": "",
		"time": {
			"start": "2026-11-11T14:10:00",
			"end": "2026-11-11T14:40:00"
		},
		"speakers": [
			{
				"id": "moshe-zioni",
				"name": "Moshe Zioni"
			}
		],
		"hall": {
			"id": "Bug Bounty Village - BSides",
			"name": "Bug Bounty Village - BSides"
		},
		"tags": [],
		"track": "Bug Bounty Village - BSides",
		"embeddings": {
			"embedTextHash": "895742283",
			"generatorVersion": "-1254175146"
		},
		"otherData": {
			"type": "talk",
			"year": 2026
		},
		"contentHash": "13h8rll-1g1np2v"
	},
	{
		"id": "bsides26-bugbounty-1440",
		"title": "When Your AI Assistant Has Root: Attacking MCP in the Wild",
		"description": "MCP has become the default way agents talk to tools - and the default way to hand an attacker a foothold. Every server an agent connects to is a new trust boundary, and most of them were built assuming the model on the other end is friendly and the user is watching. Neither is safe to assume.\n\nThis talk walks through the MCP attack surface as it actually looks in the wild: tool poisoning, confused-deputy abuse across chained servers, and the consent and OAuth gaps that let a malicious server act with the user's authority. I'll show how each class works with live examples, why current defenses miss them, and what server authors and agent builders can do today to shrink the blast radius.\n\nYou'll leave able to threat-model your own MCP deployment instead of trusting that the protocol has your back.",
		"time": {
			"start": "2026-11-11T14:40:00",
			"end": "2026-11-11T15:20:00"
		},
		"speakers": [
			{
				"id": "yuval-elbar",
				"name": "Yuval Elbar"
			}
		],
		"hall": {
			"id": "Bug Bounty Village - BSides",
			"name": "Bug Bounty Village - BSides"
		},
		"tags": [],
		"track": "Bug Bounty Village - BSides",
		"embeddings": {
			"embedTextHash": "-635232226",
			"generatorVersion": "-1254175146"
		},
		"otherData": {
			"type": "talk",
			"year": 2026
		},
		"contentHash": "1sl5gh9-1xv9tgf"
	}
]