So I guess everyone uses LLMs to think now
Web page · https://eva.ac/blog/llms
First seen by Oriel on .
AI summary
This personal essay argues that LLMs can assist skilled cybersecurity researchers but diminish the enjoyment, accountability, and learning of doing development and research yourself.
- The author avoids LLMs for blog research because investigating code, debugging, reverse engineering, and exploring tools are enjoyable, and they cannot claim credit for model-generated work.
- LLMs are powerful but require skilled guidance; an unguided model likely could not find an RCE in the Linux kernel, despite industry marketing claims.
- The author criticizes AI cybersecurity marketing, including premature proof-of-concept videos, product-linked hashes, and inconsistent narratives around php2shell.
- A Tailworm disclosure allegedly exaggerated a Tailscale privilege escalation that required existing ACL access and enabled root access on that machine, then published details without contacting Tailscale.
- The eMudhra CA private-key incident was found by a human using paranoid_crypto, which the author presents as evidence that important discoveries do not necessarily come from AI.
#llms #cybersecurity #vulnerability research #software development
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 63, latest observation ; +33 over observed period since
- Lobsters: 21, latest observation ; +19 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 85, latest observation ; +67 over observed period since
- Lobsters: 2, latest observation ; +2 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hacker News | 50330253 | https://eva.ac/blog/llms/ Read discussion |
| Lobsters | ikfsgt | https://eva.ac/blog/llms Read discussion |