Anthropic's Glasswing: 10,000+ Vulnerabilities Found in One Month, and the Patching Problem Has Never Been More Obvious

Anthropic said its Glasswing AI project found over 10,000 serious vulnerabilities in one month, exposing a growing patching gap.

Security Affairs

For folks who are thinking about locally patching open-source software to fix what they think is a bug (_especially_ if they think it's a security vulnerability), I think that's a path to https://www.xkcd.com/424/

#InfoSec #CVE #OpenSource #Mythos #Glasswing #OpenSSF

Security Holes

xkcd

#Anthropic lockert Regeln für #Mythos-KI - inside-it[.]ch

Das Unternehmen erlaubt Partnern seines Cybersecurity-Modells Mythos künftig, Erkenntnisse zu Sicherheitslücken breiter weiterzugeben.
https://www.inside-it.ch/anthropic-lockert-regeln-fuer-mythos-ki-20260519 #ArtificialIntelligence #AI #AnthropicMythos #Glasswing

Anthropic lockert Regeln für Mythos-KI

Das Unternehmen erlaubt Partnern seines Cybersecurity-Modells Mythos künftig, Erkenntnisse zu Sicherheitslücken breiter weiterzugeben.

Everyone in AI is talking about “The Mythos Effect” and Anthropic AI’s Project Mythos Glasswing. Do you have questions? What concerns should we cover? Let us know!!!
Our experts, Adrian Moua, Andrew Martin, and Liz Rice, will bring your questions into their keynotes and fireside chats at SOOCon26 On The Road’s first event in Edinburgh on 5 June. https://stateofopencon.com/edinburgh-soocon26/
#mythos #glasswing #AI #opensource #opensourceai #aiskills #stateofopencon #soocon26 #openuk #openhq #aiscotland #aimodels
OpenAI udostępnia nowe narzędzia. Ma robić to, czego ludzie nie są w stanie

Ujawniony kilka tygodni temu i udostępniony jedynie bardzo wąskiemu gronu model Claude Mythos Preview zatrząsł światem finansów i cyberbezpieczeństwa. OpenAI, znajdujące się w ostatnim czasie w def...

Business Insider Polska

Okay, here goes another AI Great Hot Take!1

AI will:

  • increase the number of discovered vulnerabilities2
  • decrease the quality of vulnerability reports3
  • increase the number of vulnerabilities4

Welcome to the #vulnslopalypse, and please take note AI added nothing of value.

1 AI.G.H.T.!
2 https://it.slashdot.org/story/26/05/11/199232/anthropics-bug-hunting-mythos-was-greatest-marketing-stunt-ever-says-curl-creator

3 https://www.computerweekly.com/feature/Vulnerability-reports-Increase-in-quantity-decrease-in-quality

4 https://www.ioactive.com/the-security-gap-in-ai-generated-code/

#genAI #Mythos #Glasswing #ProjectGlasswing #Claude

Daniel Stenberg, lead developer curlu, popisuje výsledek skenu Anthropicova modelu Mythos, toho, kolem kterého Anthropic v dubnu vyvolal rozruch tvrzením, že je „nebezpečně dobrý“ v hledání bezpečnostních chyb. Přístup curl dostal přes program Glasswing a Linux Foundation / Alpha Omega.

Výsledek: sken nad 178 000 řádky kódu ohlásil 5 „potvrzených zranitelností“. Po […]

https://zdrojak.cz/zpravicky/mythos-nasel-v-curl-zranitelnost/

271 real bugs. Real Firefox. All patched.
The scariest thing about this week's Mozilla post isn't in the bug table.
It's in the methodology section.

https://blog.ppb1701.com/better-rivets

#ai #anthropic #blog #firefox #glasswing #infosec #mozilla #mythos #security #userhostile

Better Rivets - ByteHaven - Where I ramble about bytes

Part of the ongoing Big Tech's War on Users series. Two weeks ago I wrote 271. That's The Number They're Telling You. — about the Firefox vulnerability...