buildOne report1 publisher GitHub's global advisory database lacks 195 of 618 advisories that self-hosted AI projects published
Munim, Inc.'s census of 30 self-hosted AI projects found 195 of 618 maintainer advisories missing from GitHub's global database under their own IDs. Dependabot and pip-audit read that database. The author infers OSV tools share the gap, and his direct OSV sweep is still running.
Reality
- Evidence62
- Adoption
- Not scored
- Hype gap+15
- Incentives35
- Confidence60
buildOne report1 publisher Deno's audit --fix left deno.lock untouched on a semver ReDoS with a patched version available
Deno 2.9.7's `deno audit --fix` left deno.lock byte-for-byte identical on a vulnerable semver, a dev.to author reported. The post says --ignore and --frozen-lockfile did the same, so CI needs its own check that a fix happened.
Reality
- Evidence58
- Adoption
- Not scored
- Hype gap+15
- Incentives
- Not scored
- Confidence55
buildOne report1 publisher Four September n8n advisories come back 'not found' on OSV.dev and GitHub's database
Two of n8n's three September advisory batches, at least 26 advisories in all, have no CVE IDs, a dev.to post reports. A green scanner report on a self-hosted instance says little until someone checks the version by hand.
Reality
- Evidence42
- Adoption
- Not scored
- Hype gap+20
- Incentives45
- Confidence40
buildOne report1 publisher OSV.dev returned 'not found' for five vLLM advisories that 0.31.0 fixes
vLLM published more than 30 security advisories on GitHub between 12 September and 9 October, most without a CVE ID, a dev.to author reports. The author's five OSV.dev lookups came back empty, so a clean scanner run says little about any server below 0.31.0.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+20
- Incentives40
- Confidence45
buildOne report1 publisher Open WebUI's 0.11.4 security fixes were still missing from GitHub and OSV advisory records on 11 October
At least 15 Open WebUI advisories fixed in 0.11.4, one a CVSS 8.1 session-token theft, were missing from GitHub's advisory database and OSV on 11 October. Scanners reading those databases can report 0.11.1 to 0.11.3 as clean, a dev.to post found, so operators have to check the version themselves.
Reality
- Evidence48
- Adoption
- Not scored
- Hype gap+12
- Incentives55
- Confidence42
buildOne report1 publisher Six Angular typosquats on npm claimed the real package's live version, 22.2.1
GitHub flagged six npm typosquats of Angular that claimed the live 22.2.1 version, five pulling a payload through the Wayback Machine. With the version number now correct, the defence has to sit at the package name and at whether install scripts run at all.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+10
- Incentives70
- Confidence50
CVE-2026-94545 lets attacker-supplied text in an Open Graph image reach Next.js dependencies. Vercel shipped the fix on September 22 in 16.3.6; a day later, npm audit still passed affected builds.
Reality
- Evidence62
- Adoption
- Not scored
- Hype gap+20
- Incentives35
- Confidence60
buildOne report1 publisher AWS's extended SQS and SNS clients inherit Jackson pins from a library last released in March 2024
AWS's Java extended clients for large SQS and SNS messages pull Jackson versions with seven known advisories via a library last released in March 2024. Kotlin services now have ports on Maven Central that take Jackson out of that code path.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+20
- Incentives70
- Confidence45
buildOne report1 publisher Eighteen flagged agent packages still install from npm weeks after the feeds listed them
AgentGate read the shipped code behind 30-plus OSV and GHSA flags, verified 19 npm packages as malicious, and found 18 still resolvable on publication day. Removal, not detection, is where the chain stopped.
Reality
- Evidence48
- Adoption20
- Hype gap+22
- Incentives78
- Confidence55
CVE-2026-77846 lets one attacker-supplied dot walk AshSqlite's JSON query into private fields
The Erlang Ecosystem Foundation's CNA published the entry on August 30, 2026, and ash_sqlite 0.2.18 escapes the path segments. Exposure comes down to whether your callers get to name the JSON field.
Reality
- Evidence64
- Adoption18
- Hype gap−6
- Incentives30
- Confidence58
Langflow's share button converts a saved flow into an unauthenticated code execution endpoint
CVE-2026-48519 lets anyone holding a shared Langflow playground link supply their own Python inside the build request. The exposure follows a user clicking share, so it lives in flow state rather than in server config.
Reality
- Evidence78
- Adoption
- Not scored
- Hype gap−10
- Incentives35
- Confidence66
Someone enumerated LiteLLM's key tables 36 hours after the advisory hit defender feeds
Sysdig logged a pre-auth SQL injection probe against CVE-2026-42208 from an operator who already had the Prisma schema casing right, which is the argument for putting LLM gateways on the same patch clock as edge appliances.
Reality
- Evidence64
- Adoption38
- Hype gap+12
- Incentives66
- Confidence58
Kaspersky's CVE surge has two sources, and only one of them lands in the CVE count
Kaspersky's Q2 2026 figures credit AI with both writing the bugs and finding them. The disclosures carrying working exploit code never got a CVE identifier at all.
Reality
- Evidence48
- Adoption44
- Hype gap+30
- Incentives66
- Confidence45
AI writes the Dockerfile, and the pipeline is still checking the app code
Two 2026 studies put AI-generated deployment infrastructure at worse than a coin flip, and the gates most CI pipelines run were built for application source, not config.
Reality
- Evidence58
- Adoption52
- Hype gap+12
- Incentives66
- Confidence57
buildOne report1 publisher Fabricated SQLite CVEs cleared NVD, CISA ADP and Red Hat before anyone ran the code
JFrog found that 54 of 55 advisories from one new GitHub account were machine-generated fiction. They reached enterprise scanners by the normal route, which is the problem.
Reality
- Evidence55
- Adoption45
- Hype gap+15
- Incentives60
- Confidence48