The alpha, twice a week
That was 7 min instead of 4 h 18 m.
Get the best new moments in your inbox every Wednesday and Saturday. It's free.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Lex Fridman Podcast · 7 min instead of 4 h 18 m · May 6, 2026
How Google's AI bug reports clashed with FFmpeg volunteers
Kieran Kunhya
FFmpeg Contributor, FFmpeg
Longtime FFmpeg contributor and codec engineer, and the person behind the project's widely discussed official account on X.

Lex FridmanHost · Host, Lex Fridman Podcast
Robotics and AI researcher at MIT; started the show in 2018.
About this segment
Kunhya lays out the dispute over AI-found vulnerabilities in an obscure game codec with a 90-day deadline. He argues big companies should send patches or money, not just reports.
From “FFmpeg: The Incredible Technology Behind Video on the Internet | Lex Fridman Podcast #496”, May 6, 2026. Watch the full episode
“at some point the security industry needs to realize you can't keep crying wolf like this”
Kieran Kunhyaat 1:14:42 · Lex Fridman Podcast1:14:42Takeaways
- AI reports flooded volunteers on niche codecs, all marked high priority
- Google found bugs but offered little funding or patching at first
- Crying wolf on severity leads people to ignore real security warnings
In this episode
Full episode- 55:58Why Kempf turned down tens of millions to keep VLC ad-free14:24
1:10:24How Google's AI bug reports clashed with FFmpeg volunteers6:53
Transcript
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
- 1:10:37
What happened with the Google security engineers? Just to be clear, Google are one of the biggest supporters of open source out there. They have been for a long time. It's just, I think, some things kind of went a bit overboard this time. So FFMPEG itself, and this is not like a secret, it's on the homepage, you know, it processes untrusted data. There can be security issues when you parse untrusted data, that's very normal.
- 1:11:01
But recently, what changed was Google started using AI to create security reports on an open source project. fmpeg volunteers had to deal with that they did they provided very limited funding and they even went to the media first announcing how good their ai was before the issues could be fixed and this is in the public forum yeah so reporting an issue using ai to find an issue in the code which is a security vulnerability and then reporting that publicly before you're able to fix it yeah it's announcing how good their ai is that they provided a standard 90-day industry deadline without really understanding the nature of volunteer-driven development.
- 1:11:42
In addition, this vulnerability was on an obscure 1990s game codec. The way, and let's look at it from their standpoint to begin with. Let's, you know. Yeah, can you steal me on their case? Yeah, sure. They have substantial resources working on the security of open source projects that, you know, are ubiquitous and they've used, you know, a lot of compute to do that and very expensive and very capable security researchers to do that.
- 1:12:10
And that's their viewpoint, is they are contributing by doing that. But I think that's where opinions differ. It opened up a lot of interesting fissures, I would say. It does seem that there's a portion of the security community that look at themselves a bit like building architects that never have to go to site. you know going to site is something that is a little bit beneath them the actual day-to-day construction they're there to do their security things and it's someone else's problem the security industry also kind of has a very aggressive tone towards things the language they use is extremely aggressive they use very strong language like you will get popped and to Joe Public get popped you know it means something quite bad for them it means to get hacked the way i would look at it personally is a little bit like the padlock on your home
- 1:13:04
um not everyone the padlock on your home or you know the lock on your home is there to to protect against the capabilities of of what it's there to protect it's not there to protect nuclear secrets it's not there to protect fort knox and it could be looked at that they're using ai at a level of scale to go and pick those locks and then say hey your locks your locks not secure you need to deal with this whereas actually they're the ones with resources to be able to to fix this but that seems to not be something either they'll contribute to in terms of patches or in terms of financially and at the scale of ai is kind of the issue that the the bug reports are very wordy they're very very it's almost a denial of service by ai generated bug reports on very niche codecs um and the other issue the security community has is everything is marked high priority you're going to
- 1:14:02
you know this is the most important thing in the world and you need to deal with this high high high vulnerable scary scary scary on a game codec used on one disc in 1993 yeah and that that's where the dichotomy lies going around telling everyone that their padlock's not safe well that's a hobby project of somebody the the safety of that codec is consummate to what that person thinks it's their hobby it's good that they're security analyzing it but it doesn't need a big scary warning this is a critical vulnerability um may also may recently also um So either there was another quote-unquote vulnerability it wasn't a google in this case but a filter could overflow and have an integer overflow and one of your pixels could be the wrong color and this was marked high 7.5 severity in red and at some point the security industry
- 1:14:57
needs to realize you can't keep crying wolf like this because this just leads to people you know the equivalent thereof of putting password stickers on their pc you know you can't just keep crying wolf every day and i appreciate you know that's their modus operandi is to create as much scared and fear but from the google standpoint at the end of the day they need to contribute either financially or with patches google uses ffmpeg and a scale probably you or i couldn't even contemplate millions of cpu cores and yes they contribute in areas mostly regarding their own products so vp9 av1 but in a wider sense there's a disproportionate level of contribution yes they fund students yes they fund summer of code and i think uh so alex strange is a former ffmpeg developer um i think posting in a personal capacity so he uh posted about security engineers
- 1:15:55
on hacker news uh his post reads the problem with security reports in general is security people are rampant self-promoters, in parentheses. Linus once called them something worse. Imagine you're a humble volunteer open-source developer. If a security researcher finds a bug in your code, they're going to make up a cute name for it, start a website with a logo. Google is going to give them a million-dollar bounty. They're going to go to DEF CON and get a prize and I assume some kind of secret security people orgy where everyone is dressed like they're in the matrix. Nobody is going to do any of this for you when you fix it. Basically commenting on the sort of the incentives for the different people involved and misaligned. The problem here is a disproportion of means on discovery compared to patching it, right?
- 1:16:59
And this is the biggest issue, right? And after that debacle, Google did some changes. They are now starting to send patches, which is...