cd /news/ai-products/tl-dv-breach-181k-meeting-recordings… · home topics ai-products article
[ARTICLE · art-90595] src=byteiota.com ↗ pub= topic=ai-products verified=true sentiment=↓ negative

tl;dv Breach: 181K Meeting Recordings Left Wide Open

Tl;dv, an AI meeting notetaker used by over two million people, left 181,874 meeting records fully accessible to any authenticated user for six months, including active conference IDs for live Google Meet and Teams meetings. Security researcher Bob DaHacker discovered the flaw in January 2026, reported it to the company's CTO, and received no substantive response despite repeated follow-ups through July; the vulnerability was still live when he went public on August 4. The exposed data spanned 84,312 users and 35,003 email domains, including government agencies from 23 countries and universities such as UC Berkeley and the University of Tokyo, and the company advertises SOC2 Type II compliance on its homepage.

read4 min views1 publishedAug 10, 2026
tl;dv Breach: 181K Meeting Recordings Left Wide Open
Image: Byteiota (auto-discovered)

tl;dv, the AI meeting notetaker used by over two million people, left 181,874 meeting records fully accessible to any authenticated user on its platform — for six months. A security researcher discovered the flaw in January 2026, reported it to the company’s CTO, and received no substantive response despite repeated follow-ups through July. The vulnerability was still live when the researcher went public on August 4. tl;dv, for its part, advertises SOC2 Type II compliance on its homepage.

Any Authenticated User Could Watch Your Meeting — Live #

The tl;dv security breach was not a leak of archived recordings. The exposed data included active conference IDs — the actual Google Meet and Teams room links for meetings currently in progress. The Firestore meetings

collection had no tenant isolation, meaning any logged-in tl;dv user could query every meeting record on the entire platform in real time. At any given moment, roughly 1,000 meetings were actively recording, each with a joinable link sitting in an unprotected collection.

The attack chain was straightforward: watch the Firestore collection with a real-time listener, wait for a record with status: recording

to appear, grab its conference ID, and request admission by impersonating the expected meeting bot. The researcher tested this and found it worked in approximately 80% of cases. Meeting hosts have been trained to approve bot join requests automatically. That conditioning became a vulnerability. Among the 181,874 affected records spanning 84,312 users and 35,003 email domains: government agencies from 23 countries, including Ukraine, Brazil, Israel, and the United States, plus universities including UC Berkeley and the University of Tokyo. The full technical breakdown is documented in the researcher’s disclosure writeup.

Related:[Langflow CVE-2026-9198: Patch Now or Lose Your API Keys]

The Firestore Rule That Should Have Blocked All of This #

This is a Broken Object Level Authorization (BOLA) flaw — the number one item in the OWASP API Security Top 10. The meetings

collection checked that users were authenticated but never verified they were reading their own records. Every other collection on tl;dv’s platform — users, transcripts, recordings — correctly returned 403 errors for unauthorized requests. The meetings collection was the exception, and somehow that exception made it to production.

The fix is a single additional condition in the Firestore security rule:

// Broken: checks authentication but not ownership
match /meetings/{meetingId} {
  allow read: if request.auth != null;
}

// Fixed: requires the meeting to belong to the requesting user
match /meetings/{meetingId} {
  allow read: if request.auth != null
    && resource.data.ownerId == request.auth.uid;
}

This class of mistake is not new. In 2024, Firestore misconfigurations across more than 900 websites exposed 125 million user records. Google’s own Firestore security documentation covers this exact pattern under “insecure rules.” Firebase’s low-friction approach to shipping helps startups move fast — it also helps them move fast past security rules they do not fully understand.

Why SOC2 Certification Did Not Prevent the tl;dv Breach #

tl;dv’s security page promises a “24-hour response from its security team.” The researcher waited six months. The company is SOC2 Type II certified. These two facts should inform every enterprise procurement decision involving AI SaaS tools.

SOC2 audits process controls against five trust service criteria: security, availability, processing integrity, confidentiality, and privacy. It does not require penetration testing of individual database collections. An auditor evaluating tl;dv’s processes had no obligation to query the Firestore meetings

endpoint and verify tenant isolation. The badge on the homepage signals that certain organizational processes were reviewed and found to be in place at audit time. It does not mean your meeting data is correctly access-controlled. The developer community’s reaction on Hacker News was blunt: SOC2 is “meaningless/useless” if a six-month-ignored critical vulnerability can coexist with a compliance certification.

The disclosure failure is arguably worse than the vulnerability. BOLA flaws happen. Ignoring a researcher who contacts your CTO repeatedly for half a year, while your security page promises 24-hour responses, is a governance failure. More detail on the timeline is available in the Netizen technical analysis.

Key Takeaways #

  • 181,874 tl;dv meeting records — including live, joinable conference links — were accessible to any authenticated user on the platform for at least six months.
  • The attack was not theoretical: impersonating the meeting bot to join active calls succeeded roughly 80% of the time during testing.
  • The flaw is a single missing ownership check in a Firestore security rule — BOLA, the OWASP API Security Top 10’s number one issue.
  • SOC2 certification does not test individual database endpoints. It never would have caught this.
  • If you build on Firestore, verify your security rules enforce ownership at the collection level, not just authentication.
── more in #ai-products 4 stories · sorted by recency
── more on @tl;dv 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/tl-dv-breach-181k-me…] indexed:0 read:4min 2026-08-10 ·