FreeBSD CVE-2026-4747 Log Suggests Mythos Is a Marketing Trick
jgalt212
13 points
3 comments
April 21, 2026
Related Discussions
Found 5 related stories in 112.1ms across 10,324 title embeddings via pgvector HNSW
- Mythos Discovered a CVE in Its Training Data – and That's Still Worrying chris_j · 14 pts · May 11, 2026 · 61% similar
- Has Mythos just broken the deal that kept the internet safe? jnord · 37 pts · April 10, 2026 · 59% similar
- Mythos Finds a Curl Vulnerability TangerineDream · 638 pts · May 11, 2026 · 58% similar
- Mozilla says 271 vulnerabilities found by Mythos and "almost no false positives" epistasis · 121 pts · May 07, 2026 · 57% similar
- GPT-5.5: Mythos-Like Hacking, Open to All rs_rs_rs_rs_rs · 53 pts · April 23, 2026 · 57% similar
Discussion Highlights (1 comments)
vessenes
This just feels like unbridled, blind distaste for Anthropic. Everyone's entitle to their opinion, but the blog post doesn't add much to the internet's general knowledge in my opinion. The smoking gun he wants to dig in on is that the CVE was posted 12 days before the Mythos launch, and only attributed to "Claude" not "Mythos". I can't even. The 4.7 model card talks extensively about Mythos with a bunch of benchmarks, showing significant step improvements. 12 days is a ... pretty reasonable disclosure window. Why would you tell the world your codename for your new Claude model early? There must be something else to complain about. Like, for instance, that 4.7 is not an improvement on 4.6 for my workflows. That is worth a blog post. This? ...