SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

OpenAI Launches GPT-6 Astra With Autonomous Task Execution and Cybersecurity Risks

0
·1 views

OpenAI has released GPT-6 Astra, its latest frontier model, positioning it as its most capable and aligned AI system to date. Unlike previous models that primarily provided instructions, Astra is designed to complete tasks autonomously — filling out forms, running quality checks, updating records, and writing polished documents without step-by-step guidance. The model also introduces persistent memory across coding sessions in Codex and improved judgment about when to act independently versus when to pause and seek clarification. On the computer-use benchmark OSWorld 2.0, Astra scored 72.6%, edging out Claude Opus 5's 70.2%, though all figures are self-reported by OpenAI. Notably, Astra has reached the highest cybersecurity risk tier under OpenAI's own Preparedness Framework, prompting the company to restrict its offensive capabilities at launch while permitting only defensive security applications.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Axiom CLI Tool Automates Project Setup by Detecting Dependencies Automatically

A developer has released Axiom, an open-source command-line tool designed to eliminate the manual effort involved in setting up software projects. The tool automatically detects a project's required stack, dependencies, and environment configurations, then handles installation without user intervention. Axiom is available for all major operating systems and its source code is hosted on GitHub. The developer built it to address a common frustration where mismatched dependency versions and missing tools cause projects to fail before they even run. By keeping it open source, the creator aims to let the developer community add support for additional languages, frameworks, and build systems over time.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer details how Snaptura solves DPR scaling, sticky elements, and redaction in Chrome MV3

A developer has shared technical insights from building Snaptura, a free Chrome extension for capturing DOM elements as screenshots. The article explains how device pixel ratio (DPR) must be factored into every element capture to avoid blurry or misaligned crops on high-resolution displays. It also covers how sticky and fixed-position elements are handled during full-page scroll-and-stitch captures to prevent repeated headers from appearing across folds. For privacy, the extension redacts sensitive content at the DOM level before rasterization, ensuring secret data is never written into pixels. Additional challenges discussed include managing Chrome MV3 service worker lifecycle limitations and offloading heavy tasks like GIF encoding to offscreen documents.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer Builds Fully Operational Home Security Operations Center With Custom AI

A cybersecurity professional has documented the construction of an enterprise-grade Security Operations Center (SOC) running entirely from his home, using open-source tools including Wazuh, Zeek, Suricata, and Elastic. Unlike typical home lab projects, the system runs continuously, self-monitors via a dead-man's switch, and uses tiered triage to filter out roughly 99% of alert noise before notifying the operator. A custom AI triage engine — built from scratch and combining deterministic security logic with a locally-run language model — analyzes threats without sending any data to external cloud services. The system also prioritizes remediation by highlighting vulnerabilities actively being exploited in the wild rather than generating generic vulnerability lists. The builder, who runs cybersecurity firm ByTE X Bit Technologies LLC, is publishing the full architecture and methodology as a multi-part public series.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Over Half of AI Crawler Traffic on One Site Could Not Be Verified as Genuine

A web publisher analyzing 13,491 crawler requests to their site over 30 days found that more than half came from AI vendors — including Meta, ByteDance, and Amazon — that publish no IP ranges or verification method, making it impossible to confirm the requests' true origin. A further 2,226 requests could not be verified because the publisher's local copy of vendor IP lists was outdated, while 4,318 requests came from addresses that did not match any published list. Notably, all 984 requests claiming to be Perplexity-User and all 103 claiming to be PerplexityBot had a 0% verification rate, while Applebot and GoogleOther scored 98% and 100% respectively. The analysis highlights that a user-agent string is simply a self-reported claim — anyone can spoof it with a single command — making unverified crawler statistics potentially unreliable. The findings point to a structural gap in AI crawler accountability, where site owners currently have no reliable way to confirm the identity of the majority of automated traffic they receive.

0 comments Read more at DEV Community