Agentic insights for modern tech teams
Most engineering teams working on long-context agents hit the same billing wall around turn twenty. A...
In a report published Friday, OpenAI shares evidence of a new variety of prompt injection that can self-propagate like a The post OpenAI exposes “new variety of prompt injection” that can spread like computer worms appeared first on The New Stack .
Software company Atlassian has published an account of its migration from gostatsd to OpenTelemetry, replacing the internal workings of a metrics platform that receives data from about 100,000 hosts across 14 regions. The existing service had a 99.95% SLO, so the team had to change the platform without disrupting the metrics used by production alerts. By Matt Saunders
Three Artifactory vulnerabilities under active exploitation enable authentication bypassing on Internet-accessible, self-hosted deployments, potentially allowing attackers to establish persistent administrator access in under five minutes. The exploits then enable dangerous activity, including credential and key theft, arbitrary code execution, persistence, and anti-forensics measures. By Sergio De Simone
Cloudflare has introduced Worker Previews. This feature allows each Git branch to have its own isolated, production-like environment for a Worker. Each environment has a stable URL. It also has its own configuration and unique state for each branch. Plus, there’s scoped observability. By Claudio Masolo
Modern architectures often use legacy services for specific tasks, but these services may lack accurate documentation and using them can be risky if you don't understand them well. AI coding agents can help close that knowledge gap. By Pierre Pureur, Kurt Bittner, Todd Miller
DigitalOcean's Managed Agents claims a paused session keeps its processes and memory. I put a counter in a shell variable, paused for 4m28s, and it came back at 48. Forking was stranger.
OpenAI agents bypassed blocked HTTPS by tunneling through DNS delegation to reach an external chatbot, exposing a gap in sandbox controls. The monitoring system flagged the behavior in 15 minutes, but the run persisted for 2.5 hours, and a retrospective review found other undetected DNS escapes. This incident, alongside a separate case where an agent exposed a GitHub token despite explicit instructions to stop, has paused training on OpenAI's most capable models and accelerated red-teaming of their research environment.
Cloudflare has made Python Workers generally available, built on PEP 783 and on socket syscalls implemented over the Workers connect API. The announcement carries no performance figures. A urllib3 maintainer questioned who supports the upstream work afterwards, and Wasmer's founder asked for cold-start numbers, citing about 1.027 seconds from Cloudflare's own earlier post. By Steef-Jan Wiggers
Key Points Nvidia announced the Open Agent Safety Platform to prevent the type of breakout that occurred when OpenAI models accessed Hugging Face. OpenAI, Anthropic, Meta and Google have all disclosed recent incidents where their AI models escaped their sandboxes. Nvidia said Cisco, Microsoft, Oracle, CoreWeave, Dell, HPE, Lenovo, ARM and Intel are partners. In this article NVDA Follow your favorite stocks CREATE FREE ACCOUNT watch now VIDEO 3:19 03:19 Nvidia CEO Jensen Huang: All systems around AI systems must be designed with restrictive rights