Daily Tech Digest

Daily Tech Digest — 18 November 2026

This edition: OpenAI ships GPT-6.2 Astra and Sol, partners with NATO on cyber defense, and announces new child-safety tools, AWS names new Heroes ahead of re:Invent, Looker adds native KPI/gauge/Sankey visualizations, Quad9 withdraws its DNS resolver from France over court-ordered blocking, GitHub and Microsoft pilot an AI system that ships verified fixes instead of raw vulnerability reports, and two critical CVEs (Perforce Puppet Enterprise, a CVSS 9.2 OpenJDK flaw dubbed Java4Shell) land on security radars.

Artificial Intelligence

GPT-6.2 Astra and Sol launched with improved professional outputs, coding, and computer use at lower cost

A new release introduces GPT-6.2 Astra and Sol, delivering improved professional outputs, enhanced coding capabilities, better computer-use functions, and greater efficiency at lower cost. The models are accessible across ChatGPT Work, Codex, and the API, aiming to support a range of professional and development tasks. This update marks a step forward in making advanced AI tools more cost-effective for enterprise and developer workflows.

Why it matters: The combination of higher-quality professional outputs, stronger coding assistance, and improved computer interaction at reduced cost could lower the barrier for teams adopting AI in daily workflows and development pipelines, potentially accelerating meaningful productivity gains across technical and non-technical roles.

Source: Introducing GPT-6.2 Astra and Sol

OpenAI partners with NATO to apply AI for vulnerability discovery and defense

OpenAI has announced a partnership with NATO and allied cyber agencies to apply its AI models toward identifying vulnerabilities, validating patches, and safeguarding critical infrastructure. The collaboration aims to give trusted defenders advanced tooling for automated discovery and testing of security flaws across military and civilian networks. By integrating large-language-model capabilities into defensive workflows, the program seeks to accelerate remediation cycles and improve overall resilience of essential systems.

Why it matters: The partnership signals a shift toward using generative AI for proactive defense rather than only offensive research. If the models reliably surface exploitable bugs and suggest fixes, security teams can reduce manual triage time and close gaps faster across complex, distributed environments. For technology leaders, it demonstrates a concrete pathway to embed AI-assisted vulnerability management into existing security pipelines, potentially raising the baseline security posture of critical national infrastructure.

Source: Strengthening NATO and allied cyber defenses with AI

New tools and partnerships announced to combat online child sexual abuse

A new initiative announces the availability of tools, practical guidance, and partnerships designed to prevent, detect, and respond to online child sexual exploitation and abuse. The announcement states these resources will strengthen the fight against such material, but does not specify the nature of the tools, which organizations are involved, or how the guidance will be implemented across platforms.

Why it matters: Online child sexual abuse material is a persistent security and societal threat. New tools and coordinated guidance could improve detection and removal at scale, reducing harm to victims. However, without technical details or adoption commitments, the practical impact remains uncertain. Security teams and platform engineers need to assess how these resources integrate with existing content moderation pipelines.

Source: Strengthening the fight against online child sexual abuse with new tools and support

Cloud Computing

AWS names new Heroes across four continents ahead of re:Invent

AWS has recognized a new group of AWS Heroes ahead of its re:Invent conference, highlighting individuals from North America, Oceania, and South America. The honorees include a Serverless Hero leading a user group and blogging about cloud-native architectures and generative AI, a Community Hero organizing meetups and mentoring cloud architects, and a Data Hero specializing in database reliability. Each recipient contributes to the AWS community through knowledge sharing, event organization, and technical leadership.

Why it matters: The program underscores the importance of community-driven expertise in accelerating cloud adoption. By spotlighting practitioners who translate complex topics like serverless computing and agentic AI into accessible guidance, AWS helps developers and architects worldwide build skills and confidence. The global distribution of Heroes also reflects the expanding reach of cloud-native practices across diverse regions.

Source: Four Continents, Four New AWS Heroes: Closing Out 2026

Looker adds native KPI, gauge, and Sankey visualizations in November update

Looker has announced a major visual language update rolling out in November, introducing new native chart types and dashboard controls across its platform. The update adds KPI visualizations with period-over-period delta indicators, bullet charts and radial gauges for target tracking, automatic compact number formatting, and Sankey diagrams for multi-stage flow analysis. These capabilities are built on Looker’s governed semantic foundation and integrate with its Knowledge Catalog, aiming to make data exploration and storytelling more accessible for business users.

Why it matters: The update expands Looker’s built-in visualization toolkit, reducing reliance on custom coding or external tools for common dashboard needs. KPI visualizations and compact numbers improve at-a-glance readability for executives, while Sankey diagrams provide native funnel analysis for product and marketing teams. Since these features run on Looker’s semantic layer, they maintain governed definitions across reports, which helps data teams enforce consistency. For organizations already on Looker, the rollout delivers new analytical capabilities without additional integration effort.

Source: The visual evolution of Looker: Modern design for enterprise data storytelling

Quad9 withdraws DNS resolver from France over court-ordered blocking

Swiss non-profit Quad9 announced it will withdraw its free public DNS resolver from France after a French court imposed daily financial penalties for failing to block domains linked to alleged copyright infringement. The ruling forced Quad9 to choose between crippling litigation or altering its service in ways that would undermine user privacy and security. The withdrawal deprives French users — including government agencies, schools, hospitals, and utilities — of a trusted resolver, and Quad9 lacks a direct way to notify affected parties before service ends.

Why it matters: The exit of a core internet infrastructure service from a major economy signals a growing threat to the open internet, as rightsholders push for blocking at every network layer. It erodes privacy and security for users who relied on Quad9’s encrypted, no-logging resolver, and risks disrupting critical services like healthcare and utilities that may not switch resolvers in time. The case sets a precedent that could pressure other infrastructure providers to comply with similar demands, weakening global trust in neutral DNS.

Source: Quad9’s withdrawal from France: A canary in the coal mine

Software & Systems

GitHub and Microsoft pilot AI system to fix open-source vulnerabilities

GitHub and Microsoft have launched a pilot that uses Microsoft’s Codename MDASH, a multi-model agentic scanning harness, to find and automatically propose fixes for security flaws in widely used open-source projects. The system first scans code with Microsoft’s MAI Cyber model, then an AI triage agent from the GitHub Security Lab filters out duplicates and unverified findings before any report reaches maintainers. The goal is to reduce the flood of raw vulnerability reports that currently overwhelms open-source maintainers and to deliver only actionable, verified fixes.

Why it matters: Open-source maintainers are inundated with AI-generated vulnerability reports, many of which are duplicates or false positives, slowing down real remediation. By automating triage and supplying verified fixes, the pilot aims to shift AI from a source of noise to a tool that directly reduces maintainer workload and accelerates patching of critical infrastructure. If successful, the approach could set a precedent for responsible AI-assisted security across the software supply chain.

Source: Maintainers don’t need more AI bug reports. They need fixes.

Cybersecurity

Perforce Puppet Enterprise deserialization flaw added to CISA exploited vulnerabilities list

Perforce Puppet Enterprise contains a deserialization vulnerability in its report-processing code. A client holding any CA-signed agent certificate can trigger it by submitting a specially crafted report through a PUT request, executing arbitrary code on the Puppet Server as the service user. If the certificate authority signs requests automatically without strong verification, an unauthenticated attacker on the network could obtain a valid certificate and exploit the flaw this way. The vulnerability, tracked as CVE-2026-115731, has been added to CISA’s Known Exploited Vulnerabilities catalog.

Why it matters: The flaw allows remote code execution on the Puppet Server, which manages infrastructure configuration, potentially giving an attacker control over managed systems. Because exploitation only requires a CA-signed certificate — and lax automated signing could let unauthenticated attackers obtain one — the vulnerability is practical for real-world attacks. Its inclusion in CISA’s exploited list signals active exploitation or high likelihood of abuse, urging administrators to assess their Puppet deployments and certificate authority practices.

Source: Perforce Puppet Enterprise: CVE-2026-115731 added to CISA Known Exploited Vulnerabilities

Critical “Java4Shell” flaw in OpenJDK lets attackers trigger remote code execution via malformed Base64 input

Cisco Talos is tracking a critical vulnerability in OpenJDK Java, dubbed “Java4Shell” and tracked as CVE-2026-114938, with a CVSS 4.0 score of 9.2. The flaw can lead to unauthenticated remote code execution and is triggered by sending a malformed Base64 input to an affected JVM. It affects OpenJDK Java versions 17.0.6 and later running on x86-64 CPUs with AVX-512 VBMI (Vector Byte Manipulation Instructions) support enabled — hardware commonly found across both Intel and AMD processors deployed in on-premises and cloud environments alike. Talos says the attack has already been tested successfully against a range of Java applications, and the full attack surface is likely larger as more targets are tested and confirmed. The researcher group has not observed in-the-wild exploitation yet but expects probing attempts to begin shortly after public disclosure.

Why it matters: The combination of a severe CVSS score, no authentication requirement, and a processor feature common to mainstream server hardware means a very large population of Java deployments could be affected once attackers begin probing. Because the trigger is a malformed Base64 payload rather than a complex exploit chain, reliable automated scanning for vulnerable services is likely to follow quickly. Organizations running OpenJDK 17.0.6 or later on AVX-512 VBMI-capable hardware should apply the vendor patch immediately or apply Talos’s mitigation guidance if an update isn’t yet possible.

Source: CVE-2026-114938: Critical vulnerability in OpenJDK JAVA

All Daily Tech Digest editions

Report a correction

Corrections go to the editor and are never published automatically. No account needed.