GPT 5.5

1. AI code compiles almost every time and is secure about half the time. That gap is worse in Cursor agent mode. (www.reddit.com via reddit)

   3w gpt-5cursor

   Veracode ran 100+ models this year. Two numbers from that: - it compiles basically 100% of the time - it passes security 56% of the time, and that number has not moved since last year So if you don’t specifically ask for secure code, you g…

2. Show HN: Beating GPT5.5-xhigh for Coding agent security with SLMs and IRM (harden.run via hn)

   +32 3w

   Coding agents craft arbitrary code so securing them is more complicated than red-teaming. We post trained a cyber-security small llm, changed how it reasons and supplemented our controls using program analysis techniques such as inline ref…

3. Show HN: Telem – Route agent web search across providers and inspect the traces (telem.ai via hn)

   +82 3w deepseekqwen

   TL;DR: Web search open router that routes agent web search across providers (Exa, Parallel, Tavily, Brave, SerpAPI etc.), and traces web search results with quality metrics, so you can visualize whether a bad agent run is a web search prob…

← all threads