-
product · Hacker News
From the creator of Redis; run LLM locally with ds4
The creator of Redis has released ds4, a tool for running LLMs locally, as shared on Hacker News. No specifications or performance figures are given beyond the title.
Developers choosing local inference tooling can add ds4 to their shortlist and test whether it cuts API costs and avoids sending data to outside services.
-
industry · The Verge AI · 2건
An OpenAI safety employee has quit and is sounding the alarm
David Robinson, who wrote the safety reports accompanying every major OpenAI model release, resigned this week. In an editorial in The Atlantic, he says the industry's culture is fundamentally broken, a deeper issue than adding a few rules on how models are trained.
Teams building on OpenAI models should decide whether to trust the vendor's release safety reports as-is or back them with their own evaluations, since the person who wrote those reports is now questioning the culture behind them.
-
release · The Verge AI · 2건
Meta open sources code to let you make Muse AI gadgets
Meta open sourced code that lets people build their own gadgets running its new AI agent, Muse. It says you can program an off-the-shelf ESP32 board or set up a Raspberry Pi with its SDKs, then connect Muse to displays, buttons, sensors and actuators, with project ideas including a color E Ink reminder display, an HDMI stick and a small touchscreen device.
Developers can prototype agent-driven hardware on cheap ESP32 or Raspberry Pi boards using Meta's SDKs instead of buying dedicated devices.
-
industry · Hacker News · 2건Korean coverage · GeekNews
Greg Kroah-Hartman – Security in the LLM Age [video]
Linux kernel developer Greg Kroah-Hartman gave a talk on what AI security tools achieve, where they fall short, and how maintainers respond. He noted that the 79 Linux vulnerabilities Mythos reportedly found include false positives, duplicates and already-fixed issues.
Teams adopting AI vulnerability scanners should plan staffing for triaging false positives and duplicates and for verifying and fixing bugs, since raw finding counts don't reflect real work.
-
other · Hacker News
The Four Horsemen of Agentic Coding
A post titled "The Four Horsemen of Agentic Coding" was shared on Hacker News. The title suggests it covers four problem areas in agent-based coding.
Developers adopting or running agentic coding tools can compare the four risks it names against their own tool choices and review processes.
-
industry · TechCrunch AI
Amazon responds to data center backlash, says it no longer uses NDAs
The CEO of Amazon Web Services pushed back against widespread suspicion of data centers. Amazon says it no longer uses NDAs.
If data center deals are less secret, developers weighing regions and infrastructure risk may find more public information to check.