← All reports

AI Scans Protocols Like Curl; Debate Flares Over If Machines Exceed Human Bug-Finding Prowess

AI & TechnologyApr 18, 2026score 1.004 posts · 0 replies across 4 instances
AI tools are successfully identifying vulnerabilities in protocols like curl, finding logic errors rather than textbook C language mistakes. While some funding bodies like wolfSSL and @sovtechfund are partially sponsoring fixes, the technology itself shows flaws, notably models running out of steam or varying findings per iteration. The field is split between those who see AI as an inevitable force and those who are skeptical. @[email protected] questions the premise of AI superiority, suggesting the current complaints about bug-finding are premature. Conversely, @[email protected] warns that AI’s true danger is not invention, but the amplification of every existing vulnerability. The community consensus lacks definition. Experts acknowledge AI's current impact on vulnerability discovery but are deeply divided on its practical utility. The primary fight centers on whether AI offers a demonstrable advantage over skilled human researchers, or if it merely spotlights pre-existing risks.

Key points

SUPPORT
AI found logic errors in protocols like curl, not classic C language mistakes.
[email protected] claims AIs target logic bugs specifically, noting fixes have received partial funding from wolfSSL and @sovtechfund.
SUPPORT
The danger of AI is amplifying old vulnerabilities, not creating new ones.
[email protected] cites an article suggesting 'Every Old Vulnerability Is Now an AI Vulnerability'.
OPPOSE
Claiming AI surpasses human bug-finding ability is an overreaction.
[email protected] calls the comparison premature, arguing superiority is unproven in practice.
SUPPORT
Human researchers face a psychological hurdle to keep pace with AI speed.
[email protected] suggests human detection is possible only if researchers avoid boredom.
SUPPORT
AI vulnerability findings are unreliable and inconsistent.
[email protected] notes that models can 'run out of steam' and findings differ across models.

Source posts

@[email protected]
AIs have been finding bugs and vulnerabilities in #curl for some time. Is it work to fix those? Yes. Has someone paid for this? Partially (wolfSSL and @sovtechfund) Are the AIs annoying? Yes, very. Could humans find the same bugs? Yes, if they‘d somehow avoid being bored to death through it. Was there something „heartbleed“ like? No. Were there lots of C mistakes? No, logic bugs mostly. Do AIs run out of steam? Yes. After a while a model stops finding things. Findings differ per model.
104 boosts · 105 favs · 4 replies · Apr 18, 2026
#curl
@[email protected]
🖲️ #Cybersecurity #Ciberseguridad #Ciberseguranca #Security #Seguridad #Seguranca #News #Noticia #Noticias #Tecnologia #Technology ⚫ Every Old Vulnerability Is Now an AI Vulnerability 🔗 https://www.darkreading.com/vulnerabilities-threats/every-old-vulnerability-ai-vulnerability AI's danger isn't that it's creating new bugs, it's that it's amplifying old ones.
0 boosts · 0 favs · 0 replies · Apr 17, 2026
#technology#tecnologia#noticias#noticia#news#seguranca
@[email protected]
It's a little crazy that we're essentially complaining about our ability to find and fix bugs better* and faster. (* Assuming it *is* better, on which the jury is still out) #cybersecurity #AI #mythos #codex
0 boosts · 0 favs · 0 replies · Apr 16, 2026
#cybersecurity#ai#mythos#codex
@[email protected]
Redirects for AI Training enforces canonical content Cloudflare’s Wrangler CLI has published several major versions over the past six years, each containing at least some… #NewsBeep #News #Artificialintelligence #AI #ArtificialIntelligence #Technology #UK #UnitedKingdom https://www.newsbeep.com/uk/537077/
0 boosts · 0 favs · 0 replies · Apr 18, 2026
#unitedkingdom#uk#technology#ai#artificialintelligence#news